UC-NRLF

C 5 T32 532

j^\

ff OF THE

f UNIVERSITY

MtEQNTOLO&I

BERKELEY

LIBRARY

UNIVERSITY OF
CALIFORNIA ^/EA,

-IENCE.
LIBRARY

344 PEOF. ST. GEOEGE MIVAET ON THE

The fifth rib has nouncinate process1; the flattening and expansion are less and more
dorsad ; and the tuberculum is reduced to a rudiment.

The sixth and seventh ribs have no unciform process, and no tuberculum whatever.
They end proximally in a very slender pointed process, which abuts against the end of
the supporting transverse process on the ventral aspect of the ilium.

THE STERNAL RIBS (Plate LX. fig. 4)

are all long and slender, and differ much in shape from those of Pelecanus. They
increase in length postaxiad to the fifth and sixth.

The first five expand distally and transversely somewhat, to be implanted in the
pleurosteon. The sixth and last is not expanded distally, but presents a small articular
facet at its distal end.

The fifth is slightly more than twice the length of the first.

The first and second together about equal the length of the fifth. The last two may
be grooved externally.

Sometimes2 there is a seventh sternal rib, which unites distally with the sixth, and
has no ossified vertebral rib to it.

THE STERNUM (Plate LX. figs. 4 & 5)

is relatively much longer and narrower than in Pelecanus. Its keel stretches more
preaxiad relatively.

The coracoid grooves (c) together form a more acute angle.

The median xiphoid is rounded, or may be slightly emarginate (mx). Each lateral
xiphoid (Ix) is somewhat spatulate. The sternum is not generally ankylosed to the
clavicles.

The pleurosteon is, again, very much narrower dorso-ventrally than in Pelecanus. It
presents six articular surfaces (the last very indistinct), separated by five fossae.

The ventral surface of the sternum is less convex, and its dorsal surface less concave,
than in Pelecanus, especially in its ventral half.

Its total length is nearly three times its breadth3, measured transversely at the post
axial end of keel.

The coracoid grooves are hardly separated medianly.

The costal angles (ca) are less concave on their ventral surface, and are less pointed
at their ends.

PHALACROCORAX.

The skeleton of this form is figured, under the title " Graculus" by Mr. Eyton in his
'Osteologia Avium,' plate 5 L; and a few words respecting it are to be found at p. 218.

1 Except as above. 2 E. g. in Nos. 1168, B and c, in the College of Surgeons' Museum.

3 Not always so much, e. g. in 731 A in the British Museum.

AXIAL SKELETON OF THE PELECANID^E. 345

A detailed description has been given by Brandt in the ' Memoires de 1'Academie,'
before referred to1, p. 127, with figures of the whole skeleton, the breast-bone, and a
back view of the pelvis, in pi. 2.

I have been able to make use of the following skeletons for the purpose of com-
parison:— A mounted skeleton, No. 1180 (P. carlo), and a natural skeleton, No. 1182 c
(P. Mcristatus), in the Museum of the Koyal College of Surgeons ; also a mounted
skeleton, a disarticulated natural skeleton, No. 68. 8. 16. 3 (P. carlo), a natural skeleton
from Chili, No. 76. 9. 26. 16, a disarticulated skeleton, No. 56. 11. 14. 17 (P. brasili-
ensis), and that of a small kind from the Upper Ucayle, No. 16. 3. 28. 17, all in the
British Museum.

I am also indebted to Professor Garrod for the loan of a specimen.
The total number of vertebrae, without counting the pygostyle, seems to be from
forty-six to forty-nine.

These vertebrae are subdivisible as follows — eighteen cervical, two cervico-dorsal, five
dorsal, five lumbar, two lumbo-sacral, two sacral, seven or eight sacro-caudal, and five to
eight caudal, besides the pygostyle.

There are constantly five dorsal vertebrae, and twenty cervical and cervico-dorsal
vertebrae taken together ; but there may be eighteen cervical and two cervico-dorsal
vertebrae, or seventeen cervical with three cervico-dorsals. The lumbar vertebrae are
generally five in number, but may be only four. The lumbo-sacrals may, by rare
exception, be reduced to one, or augmented to three. There may be but one vertebra
apparently distinguished as a sacral vertebra. The sacro-caudals and caudal vertebrae
taken together are generally as many as fourteen ; but they may be only thirteen, or as
many as sixteen ; and always either seven or eight of these ankylose with the pelvis, and
so come to be reckoned as sacro-caudal vertebrae.

The whole axial skeleton, when compared with that of Sula, is a somewhat yet denser
structure ; and the various ridges and processes are sharper and relatively more developed.
Median hypapophyses are not only much more developed than in Sula, but even than in
Pelecanus, especially in the anterior sacral region. Hyperapophyses are also relatively
larger than in either of the former genera. The styloid rib-like processes are free, as
in Sula, and even somewhat longer relatively, being decidedly more slender.

THE CERVICAL VERTEBRAE.

These vertebrae, when compared with those of Sula, present a very close resemblance,
but are longer relatively and more hypapophysial, with a more slender and prolonged
styloid rib (p. 35). They are more numerous than in Sula or Pelecanus ; and thus the
sudden change of form takes place over vertebras more postaxiad than in those genera.
The ATLAS has its preaxial articular surface rather less circular, broader relatively
dorsally, and narrower ventrally. There is an odontoid foramen ; but it is less decidedly

1 See above, p. 315.

\j

TRANSACTIONS

OP

THE ZOOLOGICAL SOCIETY

OF LONDON.

VOL. X.— PART 7

LONDON:

PRINTED FOE, THE SOCIETY:

SOLD AT THEIR HOUSE IN HANOVER-SQUARE:

AND BY MESSRS. LONGMANS, GREEN, READER, AND DYER, PATEBNOSTER-UOW.

August 1st, 1878.
Price 24s.

Taylor and Francis, Printer*,] [Red Lion Court. Fleet Street.

.

• t

PALEONTOLOGY

TRANSACTIONS OF THE ZOOLOGICAL SOCIETY

OF LONDON.

PART 1. (1833)

„ 2. (1834)

„ 3. (1835)

„ 4. (1835)

LUME II. (183

PART 1. (1836)

„ 2. (1838)

„ 3. (1839)

,, 4. (1840)

„ 5. (1841)

LUME III. (18^

PART 1. (1842)

„ 2. (1843)

» 3. (1844)

„ 4. (1846)

„ 5. (1848)

„ 6. (1849)

LUME IV. (IK

PART 1. (1851)

„ 2. (1852)

„ 3. (1853)

» 4. (1857)

„ 5. (1858)

„ 6. (1859)

PART 1. (1862)

„ 2. (1863)

„ 3. (1864)

„ 4. (1865)

„ 5. (1866)

LUME VI. (18(
PART 1. (1866)
„ 2. (1867)
„ 3. (1867)
„ 4. (1867)
„ 5. (1868)
„ 6. (1868)
„ 7. (1868).
„ 8. (1869)

1-1835, containing 59 Plates) .

To Follows
£ s. d.
. . Price 3 13 6

0 14 3

i. To the Public.
£ s. d.
. . . 4 18 0

. . . 0 19 0

0 15 9

. 110

156

1 14 0

6-1841, containing 71 Plates) .

. . Price 400
140

... 5 6 6
1 12 0

. . 0 15 0

100

0 14 3

... 0 19 0

42-1849, containing 63 Plates)

. . Price 386
0 14 8

... 4 11 0
... 0 19 6

0 14 8

. 0 19 6

„ 0 7 10

. 0 10 6

0 7 10

. 0 10 6

51-1862, containing 78 Plates).

. . Price 626
090

... 8 2 6
0 12 0

, 0 18 9

150

. . 0 18 0

140

. . 0 18 0

140

0 18 0

140

(1861)
(1862)

„ 0 15 9
. . , 0 18 0

... 1 1 0
140

2-1866, containing 67 Plates) .

. . Price 536
0 18 0

... 6 19 0
140

. . „ 0 18 9

150

126

1 10 0

126

1 10 0

126

1 10 0

36-1869, containing 91 Plates)

. . Price 11 5 0
1 7 0

... 15 0 0
1 IfS 0

170

1 IfJ O

170

1 Ifi ()

170

1 Hi ()

170

I]/; o

170

1 Ifi O

„ a. i \j
1 16 0

2 ft fl

170

1 16 0

[Continued

on page 3 of Wrappe>-. \

[ 315 ]

VIII. On the Axial Skeleton of the Pelecanidae. By ST. GEORGE [MivABT, F.R.S.,
F.Z.S., Professor of Biology at University College, Kensington.

Received March 27th, 1877. Read May 1st, 1877.

[PLATES LV. to LXL]

IN selecting the above-named group as the one to succeed that the axial skeleton of
which formed the subject of a previous communication to this Society1, I have thought
it well to take Pelecanus as my type and standard of comparison, first comparing it,
as regards its axial skeleton, with Struthio and the other Struthionida?, and then
comparing the other Steganopodes with it and with one another.

The osteology of Pelecanus was very briefly described by Mr. Eyton in his ' Oste-
ologia Avium,' p. 219, with a figure of the entire skeleton (pi. 7L), and with figures
of the pelvis and sternum (pi, 38. no. 2). It has been described with more detail
by Brandt, with figures of the entire skeleton, of the sternum, and pelvis (table vii.),
in the ' Mem. de 1'Acad. Imp, des Sc. de St.-Petersbourg,' 1840, 6th series, Sciences
Mathematiques, Physiques et Naturelles, tome v. seconde partie, Sciences Naturelles,
tome iii. p. 141.

PELECANUS.

The materials I have been able to make use of for the examination of this genus
are : —

The mounted specimen, No. 1167, of P. onocratalus, disarticulated skeletons
Nos. 1168 B and 1168 c, and a sternum, all in the collection of the Royal College of
Surgeons.

A disarticulated skeleton of P. rufescens, lent me by the kindness of Professor
Garrod, F.R.S.

A mounted skeleton of P, mitratus, No. 70. 3. 30. 1 ; another of P. onocratalus,
No. 527A ; two other mounted skeletons unnamed, Nos. 64, 12. 20. 1 and 67. 7. 8. 4
respectively; and a disarticulated skeleton, No. 65. 5, 3. 12, — all in the British
Museum.

The total number of vertebra? in this species seems generally to be 42, without
counting the pygostyle2.

1 Read on the 17th of November, 1874. See ante, p. 1.

2 In P. mitratus in the British Museum there are but 40 vertebras in all, there being but 3 lumbar and
5 caudal. In Nos. 67. 7. 8. 4. and 527 A there are but 41 vertebrae, there being but 3 sacro-caudals.

VOL. x. — PART vii. No. 1. — August 1st, 1878. 2 x

316 PEOF. ST. (JEOEGE MIVAET ON THE

These vertebrae may be subdivided as follows — 16 cervical, 1 cervico-dorsal1, 5 dorsal,
4 lumbar, 3 lumbo-sacral3, 3 sacral, 4 sacro-caudal3, and 6 caudal, besides the pygostyle.

The whole axial skeleton is exceedingly pneumatic, each vertebra being very light,
and having a swollen appearance with smooth surfaces and ridges, and processes but
little developed. Hyperapophyses and metapophyses are small ; and only three or four
vertebrae have conspicuous hypapophyses.

Besides the cervical vertebras, the two following ones are also generally movable.

The rest of the dorsal vertebrae and the lumbar vertebrae have their bodies ankylosed
together, and somewhat compressed laterally, but not remarkably so.

THE CERVICAL VERTEBRAE.

The cervical vertebrae unite so as to form angles with one another, open ventrad till the
seventh's junction with the eighth, which forms an angle open dorsad, as does that of the
eighth with the ninth, and so postaxiad till they unite without forming an angle at all.

The ATLAS, compared with that of Struthio, is narrower transversely in proportion to
its antero-posterior extent (Plate LV. figs. 1 & 2).

Viewed preaxially, the dorsum of the neural arch is more visible, because its pre-
axial part is more inclined ventrad. The odontoid notch is relatively deeper, and has
the minute secondary notch on the middle of its dorsal margin more marked. The
dorsal ends of the preaxial articular surface are produced more preaxiad, and rise
dorsad slightly above the level of the outward process, which they embrace. There is
no diapophysis.

Postaxially, the ventral surface of the neural arch is more visible for the reason above
given ; the odontoid notch is deeper ; and the articular surface of centrum is more
curved, but scarcely at all produced preaxially on the neural surface of the centrum.

Ventrally. there are either no catapophyses (or lateral hypapophysial processes), or
they are quite minute, while the median hypapophysis is less produced postaxiad
relatively than in Struthio.

Viewed laterally, the preaxial margin of the pedicel of the neural arch is very
slightly concave, while its postaxial margin is almost as concave as in Struthio. There
are no hyperapophyses ; and the zygapophyses are very small and inconspicuous.

The preaxial margin of the centrum is more concave than in Struthio, on account of the
greater production preaxiad of the dorsal parts of the preaxial central articular surface.

Seen dorsally, the postamal margin of the neural arch is less concave than in Struthio,
on account of the less postaxiad production of the zygapophyses.

The pedicels of the neural arch are very slender.

The AXIS is inclined more dorsad postaxially upon the atlas than in Struthio ; it is

1 With movable ribs which do not reach the sternum, and with no lateral vertebral canal.

1 The pelvic vertebrae without parapophyses.

1 Postaaeral vertebrae forming part of pelvic mass. Sometimes there are but three of these.

AXIAL SKELETON OF THE PELECANID^E. 317

also antero-posteriorly shorter, and dorso-ventrally longer, not only than in Struthio,
but even than in Urom&us, if we omit from consideration the hypapophysis of the
axis of the latter (Plate LV. fig. 3).

The odontoid process is smaller relatively than in Struthio.

Viewed preaxially, the articular surface of the centrum is more concave, and its
ventral margin more convex.

Postaxially, the articular surface of the centrum looks less dorsad and more directly
postaxiad. The transverse extent of its ventral margin considerably exceeds that of its
dorsal margin. The hypapophysis is continued more ventrad than in Struthio.

VentraUy, there may or may not be two ridges diverging postaxiad from near the
preaxial margin of the ventral surface of the centrum.

Laterally, the neural spine extends more dorsad than in Struthio, like that of
Drommis, but is longer antero-posteriorly than in the latter genus. The neural arch
extends much more postaxiad than in Struthio, so that one third of its whole antero-
posterior extent is postaxial to the postaxial end of the centrum.

The hypapophysis extends more ventrad ; its preaxial margin is concave or concavo-
convex, and inclined gently ventrad and postaxiad ; its postaxial margin is almost in
a line with the postaxial margin (as seen laterally) of centrum.

There is a small pleurapophysial lamella (at preaxial third of side of centrum)
enclosing the lateral vertebral canal, the small preaxial aperture of which is visible
when the axis is viewed laterally ; there is a slight interzygapophysial ridge, which may
be perforated ; but there is no transverse process.

Seen dorsally, more of the neural surface of the centrum is visible than in Struthio,
owing to more of the preaxial part of the neural arch having been, as it were, obliquely
cut off. The prrezygapophyses are very small ; but the postzygapophyses are nearly as
large, relatively, as in Struthio. Each is concave from without inwards, convex antero-
posteriorly ; and they look ventrad and slightly externad.

The hyperapophyses (hp] are small and inconspicuous.

The postaxial margin of the neural arch is slightly concave ; its preaxial margin is
more concave, but without any median projection preaxiad.

The THIRD VERTEBRA is much larger and half as long again as the axis. Its neural
spine is less high ; its neural arch is medianly notched postaxially ; and the much
elongated bony lamella enclosing the lateral canal is laterally prominent, and is evidently
the first appearance of a transverse process. It bears, postaxiad, a rudimentary styloid
process. It differs from that of Struthio in the greater extent of its antero-posterior
dimension compared with its transverse diameter (Plate LV. fig. 4).

Preaxially, the central articular surface is hardly to be seen, since it looks almost
entirely veu trad.

VentraUy, a hypapophysis may be seen projecting ventrad and postaxiad at the postaxial
end of centrum ; but it may be flattened on its ventral surface so as to present two lateral

2x2

318 PEOF. ST. GEOEGE MIVAET ON THE

marginal ridges, and sometimes also a median one in addition. The preaxial articular
surface of centrum is plainly visible, while immediately postaxial to it is a fossa deeper,
relatively, than in Struthio. Catapophyses begin to appear ; but conspicuous on each
side is the postaxially projecting short parapophysial styloid rib-like process.

Seen laterally, the bony lamella enclosing the lateral vertebral canal extends through
the preaxial half of vertebra, while, as before said, a rudimentary styliform parapophysial
process projects postaxiad from the ventral end of its postaxial margin. This pleur-
apophysial lamella is not continued back to reach the postzygapophysis, though a
faintly marked ridge extends so far.

The neural arch does not extend postaxiad beyond the centrum nearly so much as in
the axis, but more than in the third vertebra of Struthio or Dromceus.

Hyperapophyses plainly exist ; and metapophyses may be distinguished. The neural
spine is much less developed than in the axis.

Viewed dorsally, the lateral margins of the vertebras are more concave than in
Struthio. The preezygapophyses are convex and very elongated, and look preaxiad,
mesiad, and dorsad.

The postzygapophyses oval, look externad and ventrad, are concave transversely, and
slightly convex antero-posteriorly.

The third vertebra unites with the axis at a sharp angle, and one which opens in an
opposite direction to the angle formed by the junction of the axis with the atlas.

The preaxial end of the canal for the vertebral artery opens beside the preaxial
articular surface of centrum and beneath the prazygapophysis.

The FOURTH VERTEBRA is a little larger in most dimensions than the third. It differs
from the latter much as that of Struthio differs from its third ; but the hypapophysis has
more completely disappeared, and the neural arch more nearly approaches the preaxial
end of the centrum. The free portion of the styliform parapophysial rib is not longer
than is that of the third vertebra. There is not so much difference as to the squareness
of outline of the bone when viewed dorsally (and compared with the third vertebra) as
in Struthio, the fourth vertebra being, indeed, more antero-posteriorly elongated than
the third, but not being relatively so much more elongated in proportion to its trans-
verse breadth.

The lateral vertebral canal is much less spacious, relatively, than in Struthio or
Dromceus; and the styliform rib is much less, relatively, extended. The postaxial
margin of the neural arch is also less concave.

The preaxial articular surface of the centrum occupies less of the ventral surface of
the bone ; the catapophyses are rather more marked, and the hyperapophyses are rather
less marked than in the third vertebra.

The neural spine is longer and lower ; and, instead of any hypapophysis, the ventral
surface of the centrum is very slightly grooved medianly throughout.

AXIAL SKELETON OF THE PELECANID^E. 319

No process extends postaxially from the postaxial margin of the pleurapophysial
lamella.

The FIFTH VERTEBRA differs as much from the fourth as the corresponding one of
Struthio differs from its serial predecessor, except that the preaxial margin of the neural
arch is not more concave, and that there is no noteworthy difference as to the develop-
ment of the metapophyses.

In addition to these points, in which it differs like the fifth of Struthio from its pre-
decessor, it also differs from the fourth vertebra in that the whole ventral surface of
the centrum is deeply channelled medianly and antero-posteriorly, two parallel ridges
running postaxially from the two catapophyses (Plate LV. fig. 5).

The lateral vertebral canal is of about the same length as in the fourth vertebra ; and
therefore the part of the vertebra postaxial to it is slightly longer. The hyperapophyses
are developed to about the same degree as in the fourth vertebra; and the styliform
rib is not longer. Both this vertebra and the first differ from the corresponding
vertebrae of Struthio and Dromceus in that the postaxial margin of the neural arch is
not nearly so concave, and that the neural spine is not excavated either pre- or post-
axially. The increase in length of this vertebra over the fourth is less than in Struthio ;
it unites with its serial predecessor at a very obtuse angle.

A conspicuous foramen opens on each side of the ventral surface of the centrum, just
behind the catapophysis ; it leads into the lateral vertebral canal.

When the vertebra is viewed ventrally a wide shallow median groove is seen, bordered
on each side by a continuous catapophysial ridge. Outside the anterior third of each
catapophysial ridge is an antero-posterior (fig. 5, g) groove separating this ridge from
the postaxially extending parapophysial rib-like process.

The SIXTH VERTEBRA (Plate LV. figs. 6, 7, 8) closely resembles the fifth, and is of
nearly the same size. It differs from the fifth as does that of Struthio from its pre-
decessor, except that, when viewed dorsally, the neural surface of the centrum is not
more exposed at the preaxial end of the vertebra, and that catapophyses do not here
begin, but only increase a little in size. The subcentral antero-posterior groove is deeper
and more strongly bordered. At its preaxial end the catapophysial antero«posterior ridges
project mesiad (fig. 8, c), and draw inward (as it were) somewhat the parapophysial styloid
processes, which are here less developed than in the fifth vertebra, and are separated
from the catapophyses by a shorter, though still very conspicuous, antero-posterior
groove (fig. 8, g). The hyperapophyses are hardly smaller.

The praBzygapophyses project preaxiad of the centrum to a greater rather than a less
degree than in the fifth vertebra ; and the pre- and postaxial margins of the neural arch
are not more concave.

This vertebra differs from the fifth in the still greater production ventrad of the
lateral margins of the subcentral antero-posterior groove, and in the somewhat more

320 PKOF. ST. GEOKGE MIVAET ON THE

lateral position, and sometimes in the smaller size, of the postcatapophysial subcentral
foramen of each side (fig. 8, f).

Metapophyses and hyperapophyses are rather more marked; the proezygapophyses
look rather more preaxiad ; and the postzygapophyses look slightly more externad.

The postaxial part of the centrum extends more ventrad than in the fifth vertebra,
so that, when laterally viewed, the vertebra appears bent ventrad postaxially.

The SEVENTH VERTEBRA about equals the sixth in size. Unlike that of Struthio, it
does not differ from its serial predecessor as to the preaxiad development of the praezy-
gapophyses, since these extend preaxially beyond the centrum and parapophyses as
much as they do in the sixth vertebra, and the postzygapophyses project postaxially
beyond the centrum much more than in the sixth vertebra. It also differs from the
seventh vertebra of Dromaus, as well as from that of Stmthio, in that the rib-like
process is very much shorter, and the lateral vertebral canal is much less capacious — dif-
ferences which all obtain in the preceding vertebrae (Plate LV. figs. 9-12).

The postaxial ends of the laminae bounding the antero-posterior subcentral groove
are more or less marked as processes.

Compared with the sixth vertebra, the seventh one has a rather more prominent
neural spine ; and this is confined to the antero-posterior middle third of the neural arch. ,
It has also more marked catapophyses and a deeper subcentral groove. The postzyga-
pophyses are longer and look more externad. The hyperapophyses are smaller and
have advanced more preaxiad of the postaxial ends of the postzygapophyses.

There is a marked and sudden difference from the foregoing vertebrae in the appear-
ance (when the vertebra is viewed dorsally) of a deeply concave postaxial margin of the
neural arch. The pleurapophysial lamella is antero-posteriorly shorter ; and thus the
subcentral oval foramen of each side, leading into the vertebral-artery canal, appears
close behind its postaxial margin, and can be seen when the vertebra is viewed laterally
(fig. 9,y'). There is no free rib-like process, or but a rudimentary one. The postaxial
surface of the centrum looks almost entirely postaxiad, and very little dorsad.

The ventral margin of the postaxial surface of the centrum is more deeply notched
than heretofore.

Viewed laterally this vertebra is seen to be bent ventrad postaxially, as is the sixth
vertebra ; but the more antero-posteriorly continuous development of the catapophysial
ridges makes this less evident.

When this vertebra is viewed ventrally we see at the preaxial part of each catapo-
physial ridge two small antero-posterior grooves, the inner one of which (fig. 11, c, g)
might be taken as corresponding with the apparently similar grooves of the sixth
vertebra. Eeally, however, it is the outer groove (g) which does so, while the inner
groove is very different, not being bordered externally (as in the sixth vertebra) by the
parapophysis, but by the preaxial end of the catapophysial ridge, while it is bounded
medianly by a prominence which has no place in the sixth vertebra, but which is in

AXIAL SKELETON OF THE PELECANID^. 321

series with the side of the median ventral prominence beneath the haemal arch of more
postaxial vertebrae. The parapophysial process of the seventh vertebra is much smaller
and less conspicuous than in the sixth vertebra, and is a little antero-posterior ridge of
bone outside the postero-ventral part of the pleurapophysial lamella.

Thus the ridges bounding the median subcentral groove in more postaxial vertebrae
are catapophysial (i. e. lateral hypapophyses), the parapophysis hardly reappearing as a
small distinct process till the fifteenth vertebra.

The diapophysis does not begin to appear as a more or less distinct, though small,
postaxially projecting process till at the fourteenth vertebra.

When the seventh vertebra is viewed dorsally, the hyperapophyses are seen to have
advanced somewhat preaxiad from the ends of the postzygapophyses, which extend
much outwardly and postaxially beyond them.

The EIGHTH VERTEBRA, unlike the same vertebra in Struthio and Dromceus, presents
very great differences from its preaxial predecessor. The latter differs strikingly from
the sixth by the sudden appearance of a deep concavity at the postaxial margin of the
neural arch between the long backwardly extending postzygapophyses ; but in the
eighth vertebra other and more remarkable characters suddenly appear, some of which
are correlated with the backward extension of the postzygapophyses of the seventh
vertebra (Plate LV. figs. 13-17).

The eighth vertebra is slightly shorter than the seventh, and very slightly shorter
than the sixth, fifth, or even the fourth vertebra, though it is longer than the
third.

When the eighth vertebra is viewed dorsally the neural surface of its centrum is
more exposed than in the seventh vertebra through the oblique cutting away of the
preaxial part of its neural arch, which arch is deeply concave postaxially, even more so
than in the seventh vertebra (fig. 14).

The preaxial articular surface of the centrum (fig. 16) looks preaxiad and somewhat
dorsad (instead of almost entirely ventrad, as in the preceding vertebra), is very much con-
stricted medianly, while each lateral and expanded part is very convex dorso-ventrally,
and very slightly indeed concave transversely. The preaxial opening of the vertebral
lateral canal (v) opens above (dorsad) and postaxiad to this surface, instead of above and
beside it as in the seventh vertebra ; thus it is hardly seen when the vertebra is viewed
preaxially, but is conspicuous when the vertebra is viewed dorsally — conditions pre-
cisely the reverse of those obtaining in the seventh vertebra.

The postaxial articular surface of centrum (fig. 17) expands more laterally ventrad than
in the seventh vertebra ; and the whole surface looks postaxiad and not dorsad. Viewed
laterally the vertebra is seen to be less bent ventrad at its postaxial end than in the
seventh vertebra, and the postaxial end of the catapophysial ridge, or plate (which
bounds the subcentral groove externally), is more sharply defined as a process.

For the first time since the axis vertebra, the praezygapophyses are not carried pre-

322 PEOF. ST. GEOEGE MIVAET ON THE

axiad but dorsad, and fail to project preaxially as much as does the centrum. The
articular surface of each prsezygapophysis looks not only dorsad and mediad, but, for the
first time of all, even slightly postaxiad; it is convex antero-posteriorly, with a hardly
perceptible concavity transversely.

The postzygapophyses (i. e. the whole processes which support the articular sur-
faces, not alone those surfaces themselves) extend more directly postaxiad, and more
postaxiad of the centrum than in the seventh vertebra. The articular surface is less
broad in proportion to its length antero-posteriorly, and looks more externad and less
ventrad than in the seventh vertebra. The neural spine is much less prominent.

The metapophysis (m) appears as a slight prominence outside the prsezygapophysis.

Seen dorsally (fig. 1 4), a marked fossa is visible on each side postaxial to (and as it
were continued on from) the praezygapophysis. These fossae receive the postzygapophyses
of the seventh vertebra when it is much bent back upon the eighth vertebra. The
hyperapophyses, seen thus, also exhibit a singular change from their condition in more
preaxial vertebrae. In the seventh vertebra (as before said) they have already left the
ends of the postzygapophyses and advanced preaxiad ; but here they extend forwards
as two oblique ridges, approximating preaxially, and about as prominent as is the
neural spine.

In P. rufescens they extend forwards over the postaxial two thirds of the neural arch.

In P. onocratalus they extend yet considerably more preaxiad, and form the lateral
margins of the neural arch when this is dorsally viewed.

The pleurapophysial lamella is shorter antero-posteriorly than in the seventh ver-
tebra ; there is no styloid or distinct parapophysial process of any kind ; and the oval
subcentral foramen is more or less visible, when the vertebra is viewed laterally, post-
axial to the postaxial margin of the pleurapophysial lamella.

The lateral vertebral canal runs very obliquely, and suddenly dorsad, as it extends
preaxiad — in mai'ked contrast to its course in the seventh vertebra.

When the vertebra is viewed ventrally the inner processes of the catapophyses are
seen to have united to form a subcentral arch which hides between a quarter and a third
of the ventral surface of the centrum. The preaxial margin of this arch exhibits a
median notch with a convexity external to it on each side. Postaxially its margin is
deeply concave. Its ventral surface exhibits a blunt, median, antero-posteriorly ex-
tending prominence with a concavity on each side of it, which concavity is serially
homologous with the little antero-posteriorly extending groove on the ventral surface
of the catapophysis of the seventh vertebra.

The lateral catapophysial margins of the median subcentral groove on the ventral
surface of the centrum are still sharper and more prominent than in the seventh
vertebra, so that the groove is somewhat deepened, and it is the dorsal part of the wall
of this groove which is perforated just behind the catapophysial (or hypapophysial)
bridge by the oval foramen just mentioned.

AXIAL SKELETON OF THE PELECANID^. 323

This vertebra unites with the seventh at a marked angle, but one which opens the
reverse way from that in which the angle formed by the seventh with the sixth ver-
tebra opens. The mode of flexion here initiated is repeated in the next articulation ;
so that the eighth vertebra has both its adjacent vertebrae bent up dorsad from it,
whereas the third, fourth, fifth, and sixth vertebrae have each their two adjacent
vertebrae bent down ventrad, while the seventh vertebra has its next preaxial vertebra
bent down ventrad from it, and its next postaxial vertebra bent up dorsad from it.

The NINTH VERTEBRA (Plate LV. figs. 18-21) is very like the eighth, but slightly longer.
The preaxial articular surface of the centrum is directed somewhat more dorsad and less
preaxiad, and is a little more transversely extended. The preaxial openings of the canals
for the vertebral artery are even more plainly visible when the vertebra is viewed dorsally
(fig. 19). The articular surfaces of the praezygapophyses are directed more mediad ; and
the processes supporting these surfaces are seen to extend postaxiad as well as dorsad
when the vertebra is viewed laterally. The constriction (laterally) of the neural arch by
the fossae just postaxiad of the prsezygapophyses is more marked. The hyperapophyses
are similarly but more developed. The preaxial margin of the haemal (catapophysial)
arch extends as much preaxiad as does any part of the centrum.

The plates bounding laterally the median subcentral groove do not end in such
distinct processes, the ventral margin postaxiad of each such process sloping more
obliquely to reach the postaxial end of the centrum.

The TENTH VERTEBRA is like the ninth, but slightly longer. The hyperapophyses are
less marked, except at their postaxial ends. The neural spine stops short of the post-
axial end of the neural arch. The postzygapophyses look more externad. The haemal
arch is more antero-posteriorly extended, and the grooves on its ventral surface are
wider and deeper. The margins of the subcentral median channel subside gradually
postaxiad without there being any longer a noticeable process.

The ELEVENTH VERTEBRA, compared with the same vertebra in Struthio and Dromceus,
presents striking differences. Thus the preaxial surface of the centrum looks dorsad
instead of ventrad, and its postaxial surface looks postaxiad instead of partly dorsad.
The praezygapophysial surfaces are differently directed, and are less long, relatively, than
in Struthio ; though not than in Dromceus. There are no styliform rib processes ; but
there is a haemal arch and subcentral channel.

The whole vertebra is not so long a bone, in proportion to its other dimensions, as
in Struthio ; but it is longer relatively than in Dromceus.

This vertebra is like the tenth of Pelecanus, save that it is stouter, though not
longer. Its postaxial articular surface expands less ventrad. Its praezygapophysial
surfaces lie more open, i. e. look rather more dorsad. The hyperapophyses are less
strongly marked ; and the neural arch is more flattened dorsally. The plates bounding
the subcentral channel laterally, subside still more postaxially. The praezygapophyses
(i. e. the processes supporting the articular surfaces) extend more dorsad and less post-

VOL. x. — PART vii. No. 2. — August 1st, 1878. 2 y

324 PEOF. ST. GEORGE MIVART ON THE

axiad. The postzygapophyses scarcely extend postaxiad to the postaxial end of the
centrum ; and their articular surfaces look more externad than in the tenth vertebra.
The preaxial margin of the haemal arch may extend a little preaxiad of the preaxial
end of centrum.

The TWELFTH VERTEBRA is a little shorter and stouter than the eleventh. The
flattened neural arch is seen to expand more postaxiad when viewed dorsally ; but its
postaxial margin is somewhat less concave.

The haemal arch and its ventral grooves are wider ; and the arch itself does not
extend preaxiad beyond the preaxial end of the centrum. Both the prse- and post-
zygapophysial surfaces are somewhat broader. The former look more dorsad, and the
latter slightly more externad.

For the first time in the whole vertebral column yet examined, the postzygapophyses
fail to extend postaxiad as far as the postaxial end of the centrum.

The THIRTEENTH VERTEBRA is shorter and more massive, proportionally, than its pre-
decessor, and wider transversely (like the sixteenth vertebra of Struihio). The fossa
behind the prsezygapophysis (which has been getting less marked since the ninth
vertebra) has almost disappeared. Seen dorsally the vertebra appears more hourglass-
shaped; the neural spine begins to rise again; and a concavity begins to appeal-
behind its postaxial end; and the hyperapophyses are merely slightly marked ridges
diverging from the postaxial end of the neural spine. The haemal arch broadens, and
its ventral grooves widen (Plate LV. figs. 22-24).

The prsezygapophyses incline again slightly preaxiad and extend preaxially as far as
the centrum and even a little further.

The FOURTEENTH VERTEBRA repeats and carries further the characteristics of the pre-
ceding vertebra.

Especially marked is the lateral expansion of both pra?- and postzygapophyses ; and
now the transverse extent of the vertebra in the situation of the former begins greatly
to exceed its transverse extent as measured across the postzygapophyses. The neural
spine is more elevated. From the dorsal end of the postaxial margin of the pleurapo-
physial lamella a small inconspicuous process projects postaxiad. This is the repre-
sentative of the diapophysis.

Sometimes the haemal arch is medianly cleft, so that there seem to be two, lamellar,
closely approximating catapophyses. The canal for the vertebral artery is greatly
enlarged.

The FIFTEENTH VERTEBRA exhibits very marked changes when compared with the
fourteenth (it thus .corresponding with the eighteenth vertebra of Struthio). It is
rather shorter antero-posteriorly, rather wider transversely, with the width of its pre-
axial part greatly predominant. The pleurapophysial lamella sends back postaxiad a
marked diapophysial process (d) from the postero-outer dorsal margin of the enormously
expanded lateral vertebral canal, which has been gradually widening since the sixth

AXIAL SKELETON OF THE PELECANID,£. 325

vertebra. On the postero-outer ventral margin of the pleuropophysial lamella is a
small concave fossa. This is bounded in front by a minute tubercle, which is the
reappearance of a parapophysial process, while mediad of and behind the concave
surface is a slight ridge, which is the serial homologue of the catapophysial rid^e of
more preaxial vertebrae (Plate LV. figs. 25-29).

From the diapophysial process a ridge runs preaxiad along the dorsum of the pleur-
apophysial lamella, and slightly mesiad, ending preaxially in a marked antero-posterior
metapophysial ridge, separated from the adjacent margin of the praezygapophysial
surface by an antero-posterior shallow groove.

The haemal arch has disappeared altogether, and in its place is a median hypapo-
physis, which bends strongly preaxiad beneath the preaxial end of the centrum. On
each side of this is a wide groove, bounded externally by the parapophysial tubercle,
fossa, and ridge above noticed.

The neural arch is not cut away obliquely ; for its preaxial margin projects almost
as much preaxiad as does the centrum.

The neural spine is higher than in the fourteenth vertebra ; and the concave surface
postaxiad of it is much reduced in extent.

Not only do the prsezygapophyses again extend preaxiad (though oaly slightly so) of
the centrum, but the postzygapophyses (for the first time since the tenth vertebra)
extend considerably postaxiad of the postaxial end of the centrum.

The canal for the vertebral artery is still more capacious.

The preaxial articular surface of the centrum looks mainly ventrad (which it has not
done since the seventh vertebra) instead of preaxiad ; and the postaxial articular surface
looks dorsad instead of postaxiad, in this respect returning to the condition of the tenth
vertebra. Moreover this surface is less convex transversely than in the preceding
vertebra, and its ventral margin is less concave.

The SIXTEENTH VERTEBRA exaggerates the characters already given of the fifteenth.
The preaxial surface of the centrum is somewhat flattened out, less concave trans-
versely. The ventral margin of the postaxial articular surface of the centrum is either
very slightly concave or even slightly convex. All the zygapophysial surfaces are
smaller. The prsezygapophyses scarcely extend preaxiad of the centrum. The neural
spine is higher, and is inclined over preaxially at the summit of its preaxial end. Its
postaxial fossa is still more reduced. The dorsum of the transverse process is more
extensive, and it extends outwards further beyond the ridge, extending preaxiad and
mesiad from the end of the diapophysial process to the metapophysis, which latter is
more widely separated from the prsezygapophysis.

The ventral enclosure of the lateral vertebral canal (i. e. the pleurapophysial lamella)
is reduced to a delicate bridge of bone, very narrow antero-posteriorly. At the dorsal
end of this bridge is a slightly marked and convex (articular T) surface.

2y 2

326 PBOF. ST. GEOEGE MI V ART OX THE

The hypapophysis is much smaller than in the fifteenth vertebra, and springs from
close to the preaxial margin of the centrum.

A slightly marked ridge runs postaxiad and externad from a point mesiad of the
parapophysial root of the bony bridge above described. These lateral ventral ridges
are the serial representatives of the catapophysial ridges of more preaxial vertebrae,
and each ends postaxially in a small transverse process, which projects externad on a
line just postaxiad to the whole. For the first time since the fourth vertebra, we here
find the postzygapophyses extending postaxiad decidedly, though very little, beyond
their articular surfaces.

THE CERVICO-DORSAL VERTEBRA.

The SEVENTEENTH VERTEBRA again exaggerates the character of its serial predecessor.
The whole vertebra is antero-posteriorly shorter, and in fact somewhat smaller in
nearly all dimensions ; but the transverse processes and metapophyses (seen dorsally)
are much as in the sixteenth vertebra. The zygapophyses are a little smaller. The
neural spine is antero-posteriorly extended at its tip, whilst its postaxial fossa is so
postaxial that it is invisible when viewed dorsally. The neural spine broadens out
postaxially, exhibiting, dorsally, a median ridge and two lateral ones diverging from a
common point postaxiad.

The hyperapophyses project as two little pointed processes, one above each postzyg-
apophysis. The preaxial articular surface of the centrum widens out still more, and
looks entirely preaxiad, or only very slightly indeed ventrad. The postaxial central
surface looks postaxiad.

The hypapophysis is smaller, and the ridges on each side of its root fainter.

There is no lateral vertebral. canal. Such a canal is inclosed only by the help of the
rib ; and to support this there are two distinct articular facets — a parapophysial one a
little postaxiad of the lateral margin of the preaxial articular surface of the centrum,
and another diapophysial one beneath the postaxial end of the antero-posterior ridge of
the transverse process.

The praezygapophyses do not project so far preaxiad as does the centrum.

The seventeenth vertebra is in one line with the eighteenth.

THE DORSAL VERTEBRAE.

The EIGHTEENTH VERTEBRA is slightly smaller than the seventeenth. The preaxial
surface of the centrum is less wide transversely ; the same is the case with its postaxial
surface, which is also not at all concave at its ventral margin. The hypapophysis is
rudimentary and has no ridges on each side of its root. The transverse process is
somewhat less produced postaxially. The neural spine and its postaxial fossa are both
smaller. The prsezygapophyses fall still further short of attaining the preaxial end of
the centrum, while the postzygapophyses extend rather less postaxially beyond the

AXIAL SKELETON OF THE PELECANTD^E. 327

postaxial end of the centrum than in the seventeenth vertebra. There are similar
par- and diapophysial surfaces, but further apart.

Defects of ossification, which began to appear in the preceding vertebra between
these surfaces, on the outsides of the transverse processes, are here very conspicuous ; and
sometimes a large oval foramen (placed between the postzygapophysis and the adjacent
postaxial margin of the transverse process) leads into the substance of the haemal arch
on each side.

The NINETEENTH VEETEBEA is the first vertebra which forms, in the adult, part of the
sacral mass, being more or less ankylosed to the vertebra next succeeding. It is
slightly smaller, in all dimensions, than is the eighteenth vertebra ; but the lateral
defect of ossification is much larger and is the conspicuous lateral opening, the post-
axial intervertebral foramen being much smaller than that between the eighteenth and
nineteenth vertebrae.

The TWENTIETH and TWENTY-FIRST VERTEBRA repeat the characters of the nineteenth,
except that they are slightly smaller and more and more involved in the preaxiad
extension of the iliac ossification, which completely covers the dorsum of the twenty-
first vertebra, except that a foramen opens (dorsally to the transverse process), ex-
tending obliquely preaxiad and dorsad.

The rib-surfaces are rather less far apart in the twenty-first than in the twentieth
vertebra ; in the latter they are at their maximum of separation.

The TWENTY-SECOND VERTEBRA is still smaller, and is overlapped by the ilium itself.
The di- and parapophysial surfaces are much more approximated. The defects of
ossification are less extensive, though there is much individual variation in this respect.

The dorsal end of the transverse process sends out postaxially an arched ossification
(concave ventrally) which meets a similar but preaxially extending arched process from
the preaxial end of the dorsal part of the transverse process of the twenty-third vertebra.
Thus, when these vertebra? are viewed laterally, there is the appearance of a bony arch
extending up dorsally between and connecting them (Plate UX. figs. 1 & 3, xxn).

THE LUMBAR VERTEBRAE.

In comparing these vertebras with the so-called lumbar vertebrae of Struthio, and
with the corresponding vertebrae of most of the other Struthionidse, a striking difference
is apparent. It consists in the marked differentiation which exists in Pelecanus between
those more preaxial lumbar vertebrae which send out parapophysial transverse pro-
cesses, and those more postaxial ones which only develop ascending diapophysial
processes, and thus leave a wide interval between their vertebral centre and the aceta-
bula — i. e. a lateral acetabular or renal fossa on each side, as will be more fully noticed
in describing the ventral aspect of the pelvis.

These vertebras, therefore, are more evidentlv divisible into true lumbar vertebras and

328 PEOE. ST. GEOEGE MIVAET ON THE

lumbo-sacral vertebrae than is the case in any of the Struthionidae, except Apteryx and
Dinornis. There are three or four1 true lumbar vertebrae.

The TWENTY-THIRD VERTEBRA (or first true lumbar) resembles its serial predecessor,
except that there are no costal articular surfaces, but the transverse process is smooth
on its ventral surface, and the centrum is more flattened ventrally.

The arch formed by the junction between the postaxially extended and dorsally
curved parapophysial transverse process of this vertebra, and the similarly curved, but
preaxially extended, parapophysial transverse process of the twenty-fourth vertebra has
here become a presacral foramen.

The TWENTY-FOUBTH VERTEBRA is similar to the last described ; but the centrum is
smaller, and the parapophysial transverse process more expanded and flattened ventrally.
Its presacral foramen is smaller.

The TWENTY-FIFTH VERTEBRA is almost quite similar to the twenty-fourth.

The TWENTY-SIXTH VERTEBRA differs from the last in having its transverse process free
postaxially, and not uniting with any parapophysis of the twenty-seventh vertebra.
The postaxial margin of the transverse process forms the most ventral portion of the
preaxial margin of the great lateral acetabular fossa. The diapophysis ascends
obliquely dorsad and slightly externad to the ilium, and is generally quite distinct from
the parapophysis.

The intervertebral foramen is medianly constricted antero-posteriorly so as to
approximate to two superimposed foramina.

THE LUMBO-SACRAL VERTEBRAE.
These vertebras are three in number — namely, the TWENTY-SEVENTH, TWENTY-EIGHTH,

and TWENTY-NINTH.

These three intimately united vertebrae have the ventral surfaces of their centra
narrowing postaxiad and concave transversely, thus forming a continuous plate, antero-
posteriorly grooved (Plate LIX. fig. 3).

Their diapophyses ascend almost directly dorsad, the two more posterior on each
side (viz. those of the twenty-eighth and twenty-ninth vertebra?) being visible through
the acetabula when the pelvis is viewed laterally (Plate LIX. fig. 2).

There are no parapophyses, or only minute rudiments of such. The intervertebral
foramen behind each vertebra is more or less median, constricted antero-posteriorly.

THE SACRAL VERTEBRAE.

There are three sacral vertebras — the THIRTIETH, THIRTY-FIRST, and THIRTY-SECOND ; and
they form a series, the centra of which decrease in size postaxially, and continue on

1 Only in the specimens of P. mitratus, No. 70. 3. 30. 1, in the British Museum, have I found this number
reduced to three.

AXIAL SKELETON OF THE PELECANTD^E. 329

(though much narrowed transversely) the subcentral antero-posterior groove of the
lumbo-sacral vertebrae (Plate LIX. fig. 3).

In these three vertebrae the parapophyses become again suddenly and greatly deve-
loped (especially those of the first sacral vertebra), abutting moreover against the ilium.
The three parapophyses of each side are united together distally.

The diapophyses of these vertebrae take origin at a higher and higher level (i. e.,
more and more dorsad) as we proceed postaxially, so that in the thirty-second vertebra
they simply diverge from the lateral margin of the summit of the neural arch.

The coalesced neural spines of these and of the lumbo-sacral vertebra form a rounded
antero-posterior prominence, convex transversely, running along the middle of the
dorsum of the pelvis (Plate LIX. fig. 1).

THE SACRO-CAUDAL VERTEBEJE.

These are generally four ; but there may be only three1 of these vertebrae. They are
differentiated from those of the last category by the non-union serially and distally of
their parapophyses together, or with the diapophyses of the sacral vertebrae.

The THIRTY-THIED, THIETY-FOUETH, THIETY-FIFTH, and THIRTY-SIXTH VEETEBE^B form a

series decreasing in size postaxiad, each provided with a parapophysial process (on each
side), which ascends postaxiad and dorsad to the ilium (Plate LIX. fig. 3).

The diapophyses proceed externad from the dorsum of the neural lamina, while
the neural spines are ankylosed together antero-posteriorly as before. A foramen
is thus formed (apparent on the dorsum of the pelvis) between each serial pair of
diapophyses.

As we proceed postaxially, however, the diapophyses spring successively from a
lower (less dorsad) level as regards the neural spines ; so that generally in the last
sacro-caudal they have again completely coalesced with the parapophyses ; and thus
the antero-posteriorly extended canal between the diapophysial and parapophysial
elements, which has hitherto existed right through the vertebral column from the
axis inclusive, here ends and is finally closed.

THE CAUDAL VERTEBRAE.

•

These vertebrae may be five2 or six in number (more generally six) apart from the
terminal pygostyle (Plate LIX. fig. 6).

1 As in the specimens Nos. 527 A and 67. 7. 8. 4 in the British Museum.

2 I have only found five in the specimen lent me by Professor Garrod, and in a skeleton of P. mitratus
(No. 70. 3. 30. 1) in the British Museum. Thus in this individual (with but three lumbar vertebra;) there
are but forty vertebrae in all, apart from the pygostyle. In the specimens with only three sacro-caudals, there
were six free caudals, therefore forty-one vertebrae in all, apart from the pygostyle.

330 PEOF. ST. GEOEGE MIVAET ON THE

The THIRTY-SEVENTH VERTEBRA1 has its neural arch provided with a pair of small
prsezygapophyses for articulation with corresponding and similar postzygapophysial
processes extending from the neural arch of the last sacro-caudal vertebra.

It has a small neural spine and transverse processes inclined slightly ventrad ;
and each transverse process bears a more or less marked pit or depression on its pre-
axial surface, indicating a distinction between its diapophysial and parapophysial ele-
ments. There is no hypapophysis.

The THIRTY-EIGHTH VERTEBRA is similar to its predecessor, except that its zygapophysial
processes are a little longer, and also its transverse processes.

The THIRTY-NINTH VERTEBRA is like the thirty-eighth ; but its transverse processes are
slightly longer.

The FORTIETH VERTEBRA has slight postzygapophysial processes to its neural arch as
well as prsezygapophysial ones. It has also generally a small preaxially extended
hypapophysis at the preaxial margin of its centrum. This process may be medianly
divided, however (as in No. 70. 3. 30. 1. in the British Museum), or may be wanting
although there are six caudal vertebrae and four sacro-caudals (as in No. 64. 2. 20. 1.
in the British Museum). The transverse process is thicker, but not quite so extended
laterally as in the preceding vertebra.

The FORTY-FIRST VERTEBRA. — This is like the last ; but the neural spine is not quite so
dorsally extended ; the transverse processes are shorter, more swollen, and more antero-
posteriorly extended ; the hypapophysis is much larger and bifold, being medianly
grooved on its ventral aspect.

The FORTY-SECOND VERTEBRA is smaller in all dimensions, except that its hypapophysis
is still larger than in the vertebra last noticed, and is curved strongly preaxiad as well
as ventrad.

The PYGOSTYLE consists doubtless of many vertebra? united ; but it plainly consists
(even in the adult) of the vertebra ankylosed together. The whole structure forms a
plate antero-posteriorly and dorso-ventrally extended, but thin transversely. Its dorsal
margin is convex, its preaxial one concave and irregular ; its ventral margin is strongly
convex for the preaxial three fourths of its extent, and strongly concave for the rest ;
it meets the dorsal margin at its postaxial end, so that the whole plate ends postaxially
in a strongly curved pointed process, extending postaxiad and ventrad.

The dorsal part of the plate is, of course, formed of coalesced neurapophyses, and its
ventral part of coalesced hypapophyses. There is a more or less marked antero-
posteriorly extended ridge on each side, formed of coalesced transverse processes.
Often two superimposed foramina (one dorsal, the other ventral of the lateral ridge),
mark off a small anterior part as the ultimate visible vertebra from the much larger

1 This is, of course, the thirty-sixth when there are but three sacro-caudals.

AXIAL SKELETON OF THE PELECANID^E. 331

and terminal portion of the pygostyle. This most anterior portion represents, more or
less completely, at its preaxial end the characters of the forty-second vertebra.

THE PELVIS.

The pelvis of the adult consists of fifteen or sixteen vertebrae with the ossa
innominata. Compared with that of StrutMo and the other Struthionidee it presents
many contrasts (Plate LIX. figs. 1-3).

Preaxially viewed, the ilia form a much less steep arch (approximate dorsad at a
more obtuse angle) than in any of the Struthionidae (than even in Dromceus) ; and the
ilium, thus seen, is concave on each side. It differs from StrutMo and Ehea in that
the pubis and ischia respectively simply diverge ventrad.

Postaxially viewed, it differs from the pelvis of all the Struthionidae in the wide
(transversely extended) and dorsally flattened postaxial ends of the ilia ; the ilia not
being laterally compressed and flattened against the sides of the vertebrae included
between their most postaxial parts.

Laterally viewed, the acetabula are seen to be a little preaxiad of the antero-posterior
middle of the pelvis, measured from the preaxial end of the ilium to the postaxial end of
the pubis. The pelvis differs from that of all the Struthionidae in that (1) the ilium (il)
is less vertically expanded in front of the acetabulum, (2) that it is relatively more
vertically expanded behind the acetabulum, (3) that the ischium (i) is, at its distal end,
largely united with the ischium. Moreover the ischium extends postaxially beyond
the ilium more than in any of the Struthionidae, except StrutMo and Ehea, but about
as much as in StrutMo, though not nearly so much so as in Ehea.

The pubis (p) extends postaxiad of the ischium much more than in any of the
Struthionidse, except StrutMo, which is so exceptional in its family in this respect.
The pubis is curved as in StrutMo, but scarcely ever approximates at its tip to its
fellow of the opposite side. The obturator-foramen is long and narrow, and shaped
much as in StrutMo, except that its anterior part is not divided off, that division being
here only indicated by a rudimentary intertrochanteric process of the ischium, which
does not nearly attain the pubis (Plate LIX. fig. 2, ps).

In this respect Pelecanus differs from all the Struthionidae, except Apteryx and
Linornis. The pubis has no opposite corresponding process, or a mere rudiment of such.

There is no ilio-pectineal process, by which Pelecanus differs from all the existing
Struthionidae, if not from Dinornis also.

No process extends ventrad from the mid ventral margin of the pubis.

The antero-posterior diameter of the pelvis is rather more than twice its dorso- ventral
dimension.

There is an oval sacro-sciatic foramen. There is no such foramen in any of the
Struthionidse, except in Ehea and sometimes in Casuanus.

VOL. x. — PART vii. No. 3. — August ls£, 3878. 2z

332 PROF. ST. GEORGE MIVAET ON THE

Dorsally viewed (Plate LIX. fig. 1), the ilia are not relatively so transversely expanded
preaxiad as in any of the Struthionidse, and the lateral preacetabular margins of the ilia
are not so concave.

The antitrochanteric process (at] is situated just preaxiad of the antero-posterior middle
of the pelvis. It stands out a little less sharply than in any of the Struthionidse,
because the lateral margin of the pelvis postaxiad of it is not so concave as it is in all
that family.

The postacetahular part (il1) is much broader than in any of the Struthionidae. It is
widely flattened dorsally, with a median antero-posterior ridge, and a series of foramina
on each side of it.

The outline of the postaxial end of the pelvis differs widely from that of any of the
Struthionidae. Its general outline is deeply concave. Moreover the postaxial margin
of the ischium (which forms part of the general postaxial outline) is also deeply concave,
and therefore unlike that of any of the Struthionida3.

The pubis extends postaxiad more than in any genus of the last-named family which
has not got a pubic symphysis.

The greatest transverse diameter of the pelvis is between almost the hindermost parts
of the ischia.

The transverse diameter between the trochanteric processes widely exceeds that of
any more preaxiad part of the pelvis, in which respect it differs from the pelvis of any
existing Struthious bird.

Ventrally viewed (Plate LIX. fig. 3), the pelvis of Pelecanus differs strikingly from
those of all the living Struthionidse in the relatively great and absolutely predominating
width of its postacetabular part, also in the conspicuous lateral acetabular or renal fossae
on each side, mesiad of the acetabulum (f).

In these two points it agrees with Dinornis, though even in Dinornis the renal fossa3
are much smaller, relatively, than in Pelecanus, and are subcrescentic in form (each
crescent being convex mesiad), while in Pelecanus each fossa is subquadrate in shape, by
which it differs absolutely from all the Struthionidse. It also differs from them in that
there is a deep, though small, fossa (the supraacetabular fossa) on the ventral surface
of each ilium, just external to the expanded conjoined ends of the parapophyses of the
sacral vertebrse (fig. ?>,/'); only the innermost part of it is to be seen in a direct
ventral view. This fossa deepens at its preaxial end, and terminates preaxially by a
foramen which leads into a distinct fossa on the ventral surface of the postaxial
external part of the roof of the acetabulum. This supraacetabular fossa is marked off
by a strong ridge on its internal side.

By the presence of these fossae and foramina Pelecanus differs from all the
Struthionidse.

The ventral surface of the vertebral part of the pelvis is widest at the last lumbar
vertebra, then slowly contracts postaxiad to the third caudal.

AXIAL SKELETON OF THE PELECANID^E. 333

The parapophyses of the lumbar vertebrae form together three presacral foramina on
each side (fig. 3, sl, s2, s3). Postaxiad of this there are no parapophyses, but the wide
as well as long (though somewhat longer than wide) subquadrate renal fossa appears on
each side.

The parapophyses of the three sacral vertebra? become shorter postaxially.

Those of the sacro-caudal vertebra remain of about the same transverse extent as
those of the third sacral vertebra.

There is no fossa on the ventral surface of the ilium towards its postaxial end, any
more than there is in any of the Struthionida? ; but there is a wide surface (posterior
iliac ventral surface) of ilium extending outwards on each side between the distal ends
of the sacro-caudal parapophyses and the outer margin of the ilium (fig. 3, il').

THE ILIUM.

This bone almost extends to roof over the last dorsal vertebra but one (L e. the
twenty-first vertebra) ; an ossification of fascia connected with it continues onto two or
three vertebra more preaxial.

Seen laterally, the preacetabular part has its dorsal margin very slightly convex ; its
ventral margin is concave in even a less degree than the other is convex. The two
margins meet together preaxially. There is no ilio-pectineal process.

The ilium probably takes about the same share in forming the acetabulum that it
takes in Struthio.

The postacetabular part of the ilium decidedly exceeds in length its preacetabular
part. Seen laterally this part is at first very small (i. e. above the sacro-sciatic foramen),
but more postaxially it expands dorso-ventrally to join the ischium.

Seen dorsally this part is widest just behind the antitrochanteric process, whence it
narrows postaxially to the hinder end of the sacro-sciatic foramen, and then expands to
join the ischium.

The postaxial end of the ilium develops a large ilio-caudal process (/c), which is sepa-
rated from the sacrum by a narrow notch, and from the postaxial end of the ischium
by a much wider notch.

There is a gluteal ridge (fig. 1, gl) extending along the dorsal edge of the ilium from
the preaxial end back to the antitrochanteric process. Just behind and above this
process is a small prominence (figs. 1 & 2, st), which may be compared to the supra-
trochanteric process of the Struthionidse, and may be here termed the posttrochanteric
process. The ilium has on its ventral surface the supraacetabular fossa before described.

THE PUBIS.

This bone is long and very narrow, with a sigmoid curvature, but without any promi-
nent process. It forms no ankylosis with any other bone postaxial to its origin. It
widens (dorso-ventrally) just postaxiad to its ligamentous attachment to the ischium.

2z2

334 PEOF. ST. GEOEGE MIVAET ON THE

In shape it remarkably resembles the most posterior ribs ; so that, were we guided by
its appearance in this animal alone, the pubis would certainly be reckoned as a
pleuropophysis.

THE ISCHICM.

This is the shortest of the three bones forming each os innominatum.

It expands preaxially to form part of the acetabulum, then remains narrow to its
antero-posterior middle, and then expands dorso-ventrally to join the ilium. It does
not ankylose with the pubis postaxially, unless perhaps it may do so in aged individuals.

Its outer surface postaxiad of the ventral part of the acetabulum presents a smooth
groove, bounded ventrally at its preaxial end by a slight process (figs. 2 & 3, ps) pro-
jecting preaxiad and somewhat ventrad. This process is really the preaxial end of the
ventral margin of the ischium ; and mesiad of the margin is a long antero-posteriorly
directed groove (fig. 3, g), bounded internally (i. e. mesiad) by a rounded antero-poste-
riorly directed prominence, which runs on preaxially to the junction of the ischium
with the pubis beneath the antero-posterior middle of the acetabulum. Postaxially
this ridge runs on to the postaxial extremity of the bone. Such a groove exists in
Casuarius alone amongst the Struthionidse.

The ischium extends about as postaxiad as does the ilium, but much less so than
does the pubis.

THE RIBS.

There are eleven ribs on each side in all — namely, six vertebral ribs, and five sternal
ones.

THE VERTEBRAL RIBS.

The first vertebral rib, that of the cervico-dorsal vertebra, or seventeenth vertebra,
ends freely distad, but proximally it is provided with a distinct tuberculum and capi-
tulum. Compared with the apparently corresponding rib of Struthio, the capitulum is
much longer and the tuberculum is much shorter. The pneumatic foramen (which is
double) is situated more ventrad. There is a large unciform process, which is longer
than the length of the rib distal to it.

The second vertebral rib, or the first true rib (i. e. the first which is connected by a
sternal rib with the sternum), has again a much shorter tuberculum than in Struthio.
The unciform process is a little more dorsad in position, so that the part of the rib
ventral to it is larger than it. The rib is rather more flattened than is its serial pre-
decessor, and more bent convex preaxiad ; and its dorsal end is wider. It is a more
slender bone, relatively as well as absolutely, than in Struthio.

The third vertebral rib has again a very short tuberculum. The unciform process is
more dorsad than in the first true rib. The tuberculum and capitulum are wider apart,
and the whole bone is antero-posteriorly wider, than is the case in the second rib.

In the fourth vertebral rib the tuberculum and capitulum are not so wide apart as

AXIAL SKELETON OF THE PELECANID^E. 335

in the second true rib. The unciform process is smaller and more dorsad. This bone
is very nearly of the same length as are the two ribs preaxial to it.

The fifth vertebral rib has its tuberculum still shorter and its unciform process
smaller and more dorsad, if it is not entirely absent. This bone is not quite so curved
as is the one preceding.

The last, or sixth, vertebral rib has its tuberculum very small indeed, so that the
proximal end of this rib is characteristically small. There is no ujiciform process ; and
the bone is flattened and very little curved.

The whole series of true ribs have at first (i. e. towards the preaxial end of the series)
their main curvature situated more ventrally than at their more dorsal parts. Gradually
as we proceed postaxiad the curvature becomes more dorsad ; this is less marked in the
last but one ; and in the last there is one nearly equal curve throughout,

The true vertebral ribs are all nearly of equal length,

THE STERNAL RIBS,

These are five in number, and rapidly increase in length as we proceed postaxiad
(Plate LIX. fig. 4).

The first (attached to the first true rib and eighteenth vertebra) has two mitiute distal
articular surfaces.

The second is almost twice the length of the first, and has a, postaxial pneumatic
foramen ; in shape it resembles its predecessor.

The third is about twice the length of the first,

The fourth is not quite twice the length of the second. It is much curved, and may
be considerably flattened and expanded dorsally.

The fifth sternal rib is more than twice the length of the second, but is not quite
equal to the length of the second and third added together. It is more curved than its
predecessor, and is always singularly expanded proximally, with a postaxial process
simulating a blunt unciform process, which has, as it were, slipped down ventrad.

THE STERNUM,

The sternum differs, of course, from that of the Struthionida? in being keeled, The
keel is ankylosed to the clavicles, and extends much preaxiad of the preaxial end of the
sternum proper. The keel subsides postaxially a little behind the middle of the
sternum (Plate LIX. figs. 4 & 5).

The general surface of the sternum is more convex ventrally, and more concave
dorsally, than in Struthio. It is proportionally narrower transversely, and antero-
posteriorly longer.

The coracoid grooves (c) form together a right, if they do not even form a slightly
acute angle. They are medianly separated. Their ventral and dorsal margins extend pre-
axiad about equally; but the dorsal margin is expanded and rounded towards its inner end

336 PEOF. ST. GEORGE MIVAET ON THE

for articulation with the coracoid. In the middle of the preaxial end of the sternum a
groove proceeds obliquely ventrad and preaxiad, and may end preaxially in a deep fossa.

The costal angles (ca) are much as in Struthio. The postaxial margin of the sternum
has one median and two external xiphoid processes (mx & Ix). There is no prominence
in the wide, more or less deeply marked, concavity which extends on each side of the
median xiphoid process.

Each lateral margin of the sternum is slightly concave.

The pleurosteon is not so wide dorso-ventrally as in Struthio. It bears five articular
surfaces for the five sternal ribs. The fossa? between these surfaces are not nearly so
deep as in Struthio ; and each pair of superimposed articular surfaces are not so plainly
convex, and not even relatively so much separated from one another.

If we eliminate the keel, then the sternum of Pelecanus is much like that of Struthio ;
but the angle made by its coracoid groove is most nearly approached by Casuarius of
all the Struthionidae. The concavity on the ventral surface of each costal angle is
somewhat deeper than in Struthio.

SULA.

The skeleton of Sula is shortly described by Mr. Eyton in his ' Osteologia Avium,'
p. 220, with the figure of the entire skeleton (pi. 6 L). Of this genus I have examined
the following specimens : — a disarticulated skeleton in my own collection ; a mounted
specimen in the Museum of University College, Kensington; six specimens (1185,
1185 A, and 1186 A-P) in the Museum of the Royal College of Surgeons ; and three
specimens (Nos. 527 b, 731 a, and 779 a) in the British Museum.

The total number of vertebrae seems to be generally forty-three or forty-four, without
counting the pygostyle.

These vertebra? are subdivisible as follows : — fifteen cervical, three cervico-dorsal, and
six dorsal (the above numbers are constant in all the specimens examined) ; the
lumbar vertebras are generally three, but they may be two or four ; the lumbo-sacral
vertebrae are generally three, but they may be only two in number ; there are one or
two sacral vertebrae ; the sacro-caudal and caudal vertebrae, taken together (and
excluding the pygostyle), are generally twelve, but may, rarely, be as many as thirteen
(of these the seven last are caudal).

The whole axial skeleton, when compared with that of Pelecanus, is of a more
compact and dense texture, and less pneumatic. The various ridges and processes are
thus sharper and more marked. The styloid rib-like processes are more conspicuous,
being free and not merely forming part of the side of a subcentral groove.

Metapophyses are much more developed; but hyperapophyses are not so; while,
except in the second, third, and fourth vertebrae, where they are much larger, median
hypapophyses are still less developed and little more than rudiments, save in the
seventeenth and eighteenth vertebrae.

AXIAL SKELETON OF THE PELECANID^E. 337

Besides the cervical and cervico-dorsal vertebrae, the first two or three dorsal vertebrae
are more or less movable ; but the last dorsal ankyloses with all the more postaxial
vertebrae to the end of the sacro-caudals ; and the lumbar and last dorsal vertebrae
have their centra more compressed laterally.

The cervical vertebrae unite, as in Pelecanus, so as to form angles with one another,
open ventrad till the seventh's junction with the eighth, which forms an angle open
dorsad, as does that of the eighth with the ninth, and so on postaxiad till they unite
without forming an angle at all.

THE CERVICAL VERTEBRAE.

These vertebrae, when compared with those of Pelecanus, present differences more or
less analogous to the differences presented by those of Casuarius compared with those of
Struthio and Ehea ; for the vertebrae have become individually shorter antero-poste-
riorly, and at the same time more rugged and ridged, with relatively more prominent
processes.

The ATLAS presents a small odontoid-foramen (of) instead of a notch. The neural arch
is larger antero-posteriorly, with long postzygapophyses, generally making the post-
axial margin of the neural arch more concave. The hypapophysis is decidedly more
developed (Plate LVI. figs. 1 & 2).

The AXIS has its neural arch cut off still more obliquely preaxially ; the hyperapo-
physes are very much more developed, and the hypapophysis enormously so : it
projects ventrad and slightly, or much, postaxiad ; and its summit has two small vertical
grooves, bordered by one median and preaxial and two lateral ridges. Two minute
processes project from near the preaxial end of the ventral surface, near its middle ; and
another projects above each on each side, being placed a little more dorsad as well as
postaxiad (Plate LVI. fig. 3, ps).

The postaxial articular surface of the centrum has its dorsal margin even slightly
longer than its ventral one.

There is no pleurapophysial lamella ; but there is a conspicuous foramen on each side,
leading into the substance of the bone, and situated at the ventral part of the strongly
marked lateral fossa produced by the much stronger projection here than in Pelecanus
of the interzygapophysial ridge (?'z). The margin of the neural spine, viewed laterally,
is less convex.

The whole neural arch hardly projects so much postaxiad of the centrum as in the
genus first described.

The THIRD VERTEBRA is shorter than in Pelecanus, relatively as well as absolutely,
being no longer than the axis ; it is more concave laterally ; the neural arch is more cut
away preaxially. The hyperapophyses and hypapophysis very much longer, the latter
having preaxially still three ridges and two vertical grooves towards its extremity. The
lateral vertebral canal is much shorter, and very much more capacious, even absolutely.

338 PEOF. ST. GEORGE MIVAET ON THE

The interzygapophysial ridge is very much more prominent; and the outline of the
upper border of the neural arch is strongly concave. The prsezygapophyses look more
dorsad. There is a prominence outside the root of the rib-like styloid process (which
latter is much longer relatively than in Pelecanus), separated from the rudimentary
metapophysis (m) above it by an antero-posterior groove (Plate LVI. fig. 4).

The FOURTH VERTEBRA repeats the general characters of the third ; but the rib-like
process is longer, the metapophysis stronger, and the hypapophysis smaller (though
still large), with only a single groove on its ventral surface.

The FIFTH VERTEBRA is like the fourth, but that the still plainly parapophysial rib-like
process is slightly longer, and the metapophysis larger, with no groove on its ventral
side (Plate LVI. fig. 5).

The hypapophysis has become replaced by a more or less narrow median subcentral
groove, from the sides of the wide, more preaxial, part of which minute prominences
indicate the commencement of catapophyses or lateral hypapophyses (c).

The SIXTH VERTEBRA (Plate LVI. figs. 6-8) is rather longer, has the hyperapophyses
much more developed, the metapophyses more marked, and the subcentral groove wider.
The underpart of centrum is very convex antero-posteriorly, the vertebra being, as it
were, flexed dorsad at each end towards its ventral surface. The catapophyses are more
marked than in the fifth vertebra ; and the diverging rib-like styloid processes are still
plainly parapophysial. The vacant space between these processes and the catapophysial
ridges of the centrum corresponds with the groove between the catapophysis and the
parapophysis on each side of the centrum of the sixth vertebra of Pelecanus.

The SEVENTH VERTEBRA presents similar differences to its predecessor from the seventh
of Pelecanus, except that the postzygapophyses do not project more postaxiad, nor
is the postaxial margin of the neural arch more concave. The hyperapophyses are
advanced as in Pelecanus, but they generally form ridges diverging postaxiad from the
summit of the neural spine. The styloid parapophysial processes and catapophyses
are essentially similar to those of the sixth vertebra, but they are somewhat more
approximated (Plate LVI. figs. 9-12).

The EIGHTH VERTEBRA in Sula (Plate LVI. figs. 13-17) contrasts with its predecessor
even more than it does in Pelecanus, except that the neural arch is not cut away so much
more preaxially, that there is not so much difference as to the preaxial opening of the
lateral canal, nor in the shape and direction of the postaxial surface of centrum. The
hyperapophyses also are less marked; and the catapophyses do not form a haemal arch.
On the other hand, the metapophyses are suddenly very much developed, presenting a
rugged outstanding process, irregularly flattened behind, projecting dorsad and externad
from the dorsal part of the outside of the pleurapophysial lamella, just external to the
praezygapophysis.

The rib-like styloid processes are more slender, and closely approximated to the
Small catapophyses ; and they have become so much more ventral and median in posi.

AXIAL SKELETON OP THE PELECANID^. 339

tion compared with those of the seventh vertebra that they maybe regarded as being rather
catapophysial than parapophysial. Thus a change takes place here, resembling that de-
scribed as taking place in passing from the sixth to the seventh vertebra of Pelecanus.

The same contrasts takes place with its angles of articulation with adjacent vertebrae
as in the eighth vertebra of Pelecanus.

The NINTH VEKTEBEA, as in Pelecanus, exaggerates the characters of the eighth. Its
preaxial part is still more, as it were, pressed backwards (postaxiad) and downwards
(ventrad). In it, however, the haemal arch is still generally incomplete1. The neural
spine is a short strong process, higher and more marked than in the eighth vertebra.
Each hyperapophysis appears as a short but strongly projecting process, a line from
which to its fellow of the opposite side passes just behind the neural spine. The
metapophyses are again more developed, and project sharply dorsad externally to the
prsezygapophyses ; and each has two lateral and small prominences (tubercles) below it
on the external margin of the process, which, as a whole, is concave postaxially with a
median vertical ridge. Considerably ventrad of these tubercles is the small prominence
of the parapophysis (p), separated by a narrow groove from the long styloid rib-like
process, which is here plainly catapophysial, developing processes mesiad and preaxiad,
which may sometimes2 meet to form a haemal arch, though this is not generally
developed till in the next vertebra. Either this ninth vertebra, or else the next, has
the largest styloid ribs (Plate LVI. figs. 18-21).

The TENTH VERTEBRA is like the eighth, except that all the processes are smaller —
notably the neural spine and postzygapophyses. The metapophysis still bears the
three tubercles on its external margin, and has the vertical ridge behind. There is a
haemal arch which is somewhat antero-posteriorly grooved medianly. The parapo-
physes are generally a little less marked than in the ninth vertebra (Plate LVI. figs.
22-24).

The ELEVENTH, TWELFTH, and THIRTEENTH VERTEBRA become slightly shorter succes-
sively. They present the general characters of those of Pelecanus, due allowance being
made for the shortness of the bones. Here also, as in Pelecanus, the postzygapophyses
fail in the twelfth vertebra for the first time in the whole vertebral column to attain
the postaxial limit of the hinder end of the centrum.

The haemal arch advances preaxiad more in front of the preaxial end of the centrum
than in Pelecanus. The three lateral metapophysial tubercles are still conspicuous ;
but in the twelfth and thirteenth vertebrae the parapophyses have become undistinguish-
ably united with the side of the rib-like styloid catapophysial process.

The FOURTEENTH VERTEBRA differs from that of Pelecanus in that the haemal arch is
suddenly and entirely wanting ; and there is even no trace of a catapophysis, the
postero-ventral prominence of the bony ring enclosing the lateral vertebral canal
being purely parapophysial.

1 Complete in 527s in the British Museum. 2 As, e.g., in No. 779 A in the British Museum,

VOL. x. — PAET vii No. 4. — August 1st, 1878. 3 A

340 PEOF. ST. GEOEGE MIVAET ON THE

The FIFTEENTH VERTEBKA (Plate LVI. figs. 25-28) is like of Pelecanus ; but there is no
hypapophysis, only a slight median ridge, which also occurs in the fourteenth vertebra.
The preaxial surface of centrum does not look so much ventrad ; and there is no di-
stinct articular surface at the ventral part of the outer surface of the pleurapophysial
lamella. On each side there is a small foramen in the antero-posterior middle of the
body, leading into the centrum from just above the ventral margin of the body.

CERVICO-DORSAL VERTEBE^E.

The SIXTEENTH VEETEBBA has the general character of the corresponding one of
Pelecanus; but the hypapophysis is still rudimentary. There is no pleurapophysial
lamella ; but a sharp pointed process projects ventrad from the ventral surface of the
distal end of the transverse process, and may be connected with the parapophyses by a
minute osseous spiculum forming a rudimentary pleurapophysial lamella, while imme-
diately postaxial to this small ventrally extending process is a subtriangular concave
articular surface. The lateral foramen entering the centrum is very conspicuous.

The SEVENTEENTH VEETEBBA is generally like that of Pelecanus ; but the hypapophysis,
though a little larger than hitherto, is still very small. The lateral foramen leading
into the centrum is smaller than in the sixteenth vertebra, and but very little larger
than in the fifteenth.

The EIGHTEENTH VEETEBBA has a long rib ending freely, with an unciform process about
as long as the length of that part of the rib which is distal to it.

DOBSAL VERTEBRAE.

The NINETEENTH, TWENTIETH, and TWENTY-FIBS! have the general characters of the
corresponding ones of Pelecanus, except that they are the first three dorsal vertebrae.

The TWENTY-SECOND VEETEBEA is less completely over-shadowed by the pelvis than
in Pelecanus. It has a rib which is attached very slightly to the parapophysis, against
which it abuts by its very slender proximal end. The parapophysis is almost on a level
with the root of the diapophysis (Plate LX. fig. 3, xxn).

The TWENTY-THIRD VERTEBRA is similar, and differs from that of Pelecanus in that it
has a long and slender rib.

It forms part of the ankylosed pelvic mass. The diapophysis and parapophysis are
here united into a single " transverse process," and receive the slender proximal end of
the fifth thoracic rib.

The TWENTY-FOURTH VERTEBRA is similar, but smaller, more compressed, and with a
smaller transverse process.

THE LUMBAR VERTBER<E.

The TWENTY-FIFTH, TWENTY-SIXTH, and TWENTY-SEVENTH VERTEBRA resemble the lumbar
vertebrae (the twenty-third, twenty-fourth, twenty-fifth, and twenty-sixth) of Pelecanus,
except that generally these are only three in number. There are but two presacral fora-
mina on each side, enclosed between their distally expanding parapophyses. Sometimes,

AXIAL SKELETON OP THE PELECANID^E. 341

however (as in No. 1186 D in the College of Surgeons' Museum), there are four lumbar
vertebrae and three presacral foramina. The more postaxial foramen also is smaller
compared with its serial predecessor than is the most postaxial one of Pelecanus.

THE LUMBO-SACBAL VERTEBRAE.

These are generally three in number (normally the TWENTY-EIGHTH, TWENTY-NINTH,
and THIRTIETH), and resemble those of Pelecanus in most of their characters. Some-
times there are but two such vertebrae.

THE SACRAL VERTEBRAE.

These are but one or two1 in number (from the THIRTIETH, the THIRTY-FIRST and
THIRTY-SECOND VERTEBRA) ; and, unlike Pelecanus, the parapophysis of the second (when
there are two) may be longer and much stronger than that of the first. Their diapo-
physes arise at a very high (dorsal) level.

THE SACRO-CAUDAL VERTEBRAE.
They are from four to six in number, including the THIRTY-THIRD, THIRTY-FOURTH,

THIRTY-FIFTH, THIRTY-SIXTH, and THIRTY-SEVENTH VERTEBRAE (Plate LX. fig. 3).

They differ strikingly from those of Pelecanus in that they have their parapophyses
quite suddenly bent up, so that they are as much differentiated off from the sacral
vertebras as are the lumbo-sacrals in front ; and there is consequently a fossa on each
side behind the sacral parapophyses. The parapophyses gradually descend postaxially,
conjoined with the diapophyses, and being more and more inclined postaxiad.

The last three interparapophysial spaces open widely on the dorsum of the pelvis.

THE CAUDAL VERTEBRAE.

These are seven2, apart from the pygostyle, in number. Unlike those of Pelecanus
they increase in breadth considerably and gradually postaxiad.

A hypapophysis does not appear till the fifth or sixth, and it is longer on the
seventh.

The transverse processes are longest in the sixth caudal vertebra. The pygostyle is
long, prismatic, with a sharp dorsal ridge ; it is ventrally flattened ; and its ventral
margin is very slightly concave (Plate LIX. fig. 7).

THE PELVIS.

This complex structure, including fifteen vertebrae in its extended ankyloses, pretty
closely resembles the pelvis of Pelecanus (Plate LX. figs. 1-3).

Laterally viewed, however, the opening of the acetabulum is smaller, the sacro-sciatic
foramen is longer, the obturator-foramen narrower, and the pubis more poslaxially
extended (Plate LX. fig. 2).

1 Very rarely three, as in No. 527 6 in the British Museum.
a There are eight in No. 527 B in the British Museum.

3A2

342 PKOF. ST. GEOKGE MIVAET ON THE

The posterior margin of the ischium looks more dorsad, and develops a much smaller,
submedian, truncated or pointed ilio-caudal process1. The sacro-iliac ankylosis is rela-
tively as well as absolutely shorter. There is a faint indication of an ilio-pectineal
process.

Viewed dorsally, the preacetabular iliac surfaces are more concave and very much
narrower, so that the sacro-sciatic foramina come plainly into view. The external margin
of each ischium is more concave (Plate LX. fig. 1).

The shape of the postaxial margin of the pelvis presents a still deeper excavation,
the margins of which (antero-posteriorly considered) are on each side first convex
mesiad, and then mesially concave. From the second preaxiad fifth of the convexity
(which convexity is formed by the ischium) a process (the ilio-caudal spine) projects
mesiad and postaxiad (figs. 1, 2, 3, Ic).

The transverse diameter between the antitrochanteric processes but little exceeds
that of the preaxial part of the ilia, while in Pelecanus it is nearly double.

Ventrally examined, the pelvis of Sula presents some remarkable differences from that
of Pelecanus, in spite of the general similarity between the two (Plate LX. fig. 3).

In the first place the ventral surface of the most preaxial part of each ilium extends
much more out laterally beyond the ends of the vertebral transverse processes. There
are but two presacral foramina on each side2 (fig. 3, sl & s2).

The renal or lateral acetabular fossse (/) are relatively, as well as absolutely, smaller.
There are but two3 sacral parapophyses on each side ; and behind these there is on each
side an elongated fossa produced by the sudden dorsad flexion of the sacro-caudal
parapophyses. Did these processes extend out as in Pelecanus, they would reach
almost, if not quite, to the margin of the ilium. The postacetabular parts of the
ilia are so narrow that the sacro-sciatic foramina come plainly into view, there being
no wide posterior iliac ventral surface as in Pelecanus ; also the surface of the sacro-iliac
ankylosis is very small.

There is only a very small opening on each side dorsad of the expanded ends of the
sacral parapophyses ; and this seems only to lead into the substance of the bone ; for
there is no supraacetabular fossa.

There is an indication of an ilio-pectineal process (Ip) ; and the most preaxial part of
the obturator-foramen appears from this point of view to be (though it is not) separated
off by the here more marked process of the ischium.

1 This process is very prolonged and pointed in S. augtralis (1186 A), S.fuscat (1186 B), and 8. piscator
(1186 c, College of Surgeons); also in a species from Chili (No. 779 A in the British Museum).

2 As has heen said, there may be three.

* There may be three or only one, as before stated. Only in S. capensis (No. 527 in the British Museum)
have I found three such,

AXIAL SKELETON OF THE PELECANID^E. 343

THE ILIUM.

This bone extends over the last and part of the penultimate dorsal vertebrae ; and
ossification of fascia extends to the preaxial end of the last dorsal vertebra but two, i. e.
the twenty-second vertebra.

Seen laterally, the preacetabular part is less convex dorsally, convex preaxially, and
concave postaxially at its ventral margin.

There is a faint indication of an ilio-pectineal process, and a rudimentary posttro?
chanteric process (figs. 1 & 2, st). There are marked antitrochanteric processes (at).

The postacetabular part of the ilium has its dorsal margin convex, sloping down
ventrad behind the last sacro-caudal, and giving out a short, postaxiad, slightly dorsad
process (the ilio-caudal process), which, however, may belong rather to the ischium.
This process is greatly extended, and becomes a long pointed process, in some species,
as before stated. It is never so wide and massive as in Pelecanus. The dorsal margin
of the sacro-sciatic foramen is more concave ; and the foramen is longer.

THE PUBIS

is like that of Pelecanus, but extends even more postaxiad relatively, and its preaxial
part is less bent and convex ventrad. It is somewhat incurved mediad and very much so
distally.

THE ISCHIUM

has its external smooth antero-posterior groove rather deeper and narrower than in
Pelecanus, but quite distinct ; the process (fig. 2, ps) in which its inferior boundary ends
preaxially is rather more prominent and distinct. The internal groove is longer, nar-
rower, and in a sigmoid curve (Plate LX. fig. 3, g).

The postaxial expanded part of the ischium is smaller ; and its postaxial margin
slopes more preaxiad and ventrad, and probably develops the ilio-caudal process pro-
jecting postaxially at about where it joins the ilium.

The ventral surface of the ilio-ischium postaxiad to the sacro-sciatic foramen (mediad
of the ridge bounding the ventral ischiatic antero-posterior groove medianly) is a smooth
expanse, as also in Pelecanus ; but it is here less concave.

THE RIBS,
There are seven long vertebral ribs, and six or seven sternal ones.

THE VERTEBRAL RIBS.

They increase slightly in length to the fifth or sixth, and expand generally somewhat
distally, especially the fourth, fifth, and sixth. The first four1 have long uncinate
processes, which arise more dorsad as we proceed postaxiad. They are a good deal
flattened and expanded towards the middle of their superuncinate part.

1 Five sometimes — e.g. in 1186s, c, and D, College of Surgeons.

344 PEOF. ST. GEOEGE MIVAET ON THE

The fifth rib has nouncinate process1; the flattening and expansion are less and more
dorsad ; and the tuberculum is reduced to a rudiment.

The sixth and seventh ribs have no unciform process, and no tuberculum whatever.
They end proximally in a very slender pointed process, which abuts against the end of
the supporting transverse process on the ventral aspect of the ilium.

THE STEKNAL RIBS (Plate LX. fig. 4)

are all long and slender, and differ much in shape from those of Pelecanus. They
increase in length postaxiad to the fifth and sixth.

The first five expand distally and transversely somewhat, to be implanted in the
pleurosteon. The sixth and last is not expanded distally, but presents a small articular
facet at its distal end.

The fifth is slightly more than twice the length of the first.

The first and second together about equal the length of the fifth. The last two may
be grooved externally.

Sometimes2 there is a seventh sternal rib, which unites distally with the sixth, and
has no ossified vertebral rib to it.

THE STERNUM (Plate LX. figs. 4 & 5)

is relatively much longer and narrower than in Pelecanus. Its keel stretches more
preaxiad relatively.

The coracoid grooves (c) together form a more acute angle.

The median xiphoid is rounded, or may be slightly emarginate (mx}. Each lateral
xiphoid (Ix) is somewhat spatulate. The sternum is not generally ankylosed to the
clavicles.

The pleurosteon is, again, very much narrower dorso-ventrally than in Pelecanus. It
presents six articular surfaces (the last very indistinct), separated by five fossae.

The ventral surface of the sternum is less convex, and its dorsal surface less concave,
than in Pelecanus, especially in its ventral half.

Its total length is nearly three times its breadth3, measured transversely at the post
axial end of keel.

The coracoid grooves are hardly separated medianly.

The costal angles (ca) are less concave on their ventral surface, and are less pointed
at their ends.

PHALACROCORAX.

The skeleton of this form is figured, under the title " Graculus," by Mr. Eyton in his
'Osteologia Avium,' plate 5 L; and a few words respecting it are to be found at p. 218.

1 Except as above. 2 E. g. in Nos. 1168, B and c, in the College of Surgeons' Museum.

* Not always so much, e. g. in 731 A in the British Museum.

AXIAL SKELETON OP THE PELECANTD.E. 345

A detailed description has been given by Brandt in the ' Memoires de 1'Academie,'
before referred to1, p. 127, with figures of the whole skeleton, the breast-bone, and a
back view of the pelvis, in pi. 2.

I have been able to make use of the following skeletons for the purpose of com-
parison:— A mounted skeleton, No. 1180 (P. carbo), and a natural skeleton, No. 1182 c
(P. bicristatus), in the Museum of the Royal College of Surgeons ; also a mounted
skeleton, a disarticulated natural skeleton, No. 68. 8. 16. 3 (P. carbo), a natural skeleton
from Chili, No. 76. 9. 26. 16, a disarticulated skeleton, No. 56. 11. 14. 17 (P, brasHi*
ensis), and that of a small kind from the Upper Ucayle, No. 16. 3. 28. 17, all in the
British Museum.

I am also indebted to Professor Garrod for the loan of a specimen.

The total number of vertebrae, without counting the pygostyle, seems to be from
forty-six to forty-nine.

These vertebrae are subdivisible as follows — -eighteen cervical, two cervico-dorsal, five
dorsal, five lumbar, two lumbo-sacral, two sacral, seven or eight sacro-caudal, and five to
eight caudal, besides the pygostyle.

There are constantly five dorsal vertebrae, and twenty cervical and cervico-dorsal
vertebrae taken together ; but there may be eighteen cervical and two cervico-dorsal
vertebrae, or seventeen cervical with three cervico-dorsals. The lumbar vertebrae are
generally five in number, but may be only four. The lumbo-sacrals may, by rare
exception, be reduced to one, or augmented to three. There may be but one vertebra
apparently distinguished as a sacral vertebra. The sacro-caudals and caudal vertebrae
taken together are generally as many as fourteen ; but they may be only thirteen, or as
many as sixteen ; and always either seven or eight of these ankylose with the pelvis, and
so come to be reckoned as sacro-caudal vertebras.

The whole axial skeleton, when compared with that of Sula, is a somewhat yet denser
structure ; and the various ridges and processes are sharper and relatively more developed.
Median hypapophyses are not only much more developed than in Sula, but even than in
Pelecanus, especially in the anterior sacral region. Hyperapophyses are also relatively
larger than in either of the former genera. The styloid rib-like processes are free, as
in Sula, and even somewhat longer relatively, being decidedly more slender.

THE CERVICAL VERTEBRAE.

These vertebrae, when compared with those of Sula, present a very close resemblance,
but are longer relatively and more hypapophysial, with a more slender and prolonged
styloid rib (p. 35). They are more numerous than in Sula or Pelecanus • and thus the
sudden change of form takes place over vertebrae more postaxiad than in those genera.
The ATLAS has its preaxial articular surface rather less circular, broader relatively
dorsally, and narrower ventrally. There is an odontoid foramen ; but it is less decidedly

1 See above, p. 315.

346 PEOF. ST. GEOBGE MIVAET ON THE

enclosed dorsally. The hyperapophyses are still longer and more diverging, and con-
sequently the postaxial margin of the neural arch more concave (Plate LVII. figs. 1-3).

There are three hypapophysial processes : — one long median, laterally compressed ;
and two small pointed lateral ones, which are really rudimentary styloid ribs.

The AXIS (Plate LVII. figs. 4 & 5) has its hyperapophyses somewhat less developed
than in Sula ; its hypapophysis more convex preaxially, curved so that its tip is almost
dorsad ; and instead of the two minute pointed lateral processes in front of the hypapo-
physis there are two longer processes connected medianly with the hypapophysis — in fact,
two small, but distinct, rib-like styloid processes (ps) which, considered by themselves,
might be either catapophysial or parapophysial, but which should rather be considered
parapophysial, as they are evidently in series with the parapophysial processes which
succeed them postaxially. There is no lateral foramen leading into the centrum.

The THIED VERTEBRA (Plate LVII. figs. 6 & 7) is longer in proportion to its breadth
than in Sula. The prsezygapophyses look somewhat less directly dorsad. The spine
and hyperapophyses are sharper and more developed. The postzygapophyses are more
concave and directed more postaxiad. The styloid rib-like processes are smaller relatively,
as well as absolutely, than in Sula, though they have greatly increased compared with
those of the axis.

The FOURTH VERTEBRA differs much as does the third, except that the hypapophysis
is less (is much less) dorso-ventrally developed, though more equally so for a greater
part of its length antero-posteriorly.

The FIFTH VERTEBRA is not grooved beneath, but has an antero-posteriorly elongated
hypapophysial ridge but little prominent.

The SIXTH VERTEBRA is also ridged beneath instead of grooved ; and the neural spine
and especially the hyperapophyses and metapophyses are generally more sharply pro-
minent ; but the styloid rib-like processes are hardly less developed relatively.

The SEVENTH VERTEBRA (Plate LVII. figs. 8-11) is grooved beneath, for the first time
(as is the fifth of Sula), and (as also that fifth) rudimentary ; catapophyses begin to
appear at the preaxial part of the groove. It differs from the seventh vertebra of Sula
in that it is grooved beneath narrowly instead of broadly ; otherwise it differs from the
vertebra of the same number in Sula as does the sixth ; but the fossa at the preaxial
part of its ventral surface is deeper. The styloid rib starts postaxiad from the outer side
of the antero-posteriorly grooved lamella of bone which exists on each side of the pre-
axial end of the median subcentral groove, and is plainly parapophysial, as it is in Sula.

The EIGHTH VERTEBRA (Plate LVII. figs. 12-16) differs from the eighth of Sula in
not being pressed back preaxially, and in not diverging widely postaxially ; and the me-
tapophyses are not here suddenly increased. The praezygapophyses still project decidedly
preaxiad of the centrum ; and the postzygapophyses1 diverge but little more than in the

1 These have here their articular surface more elongated and quite extending to the postaxial ends of the
processes, which they do not in the seventh vertebra, and hardly in the seventh of Sula. In Pelecanus they

AXIAL SKELETON OF THE PELECANID^E. 347

seventh vertebra, though the hyperapophyses have come to join and diverge from the
neural spines. Thus this vertebra is like the seventh ofPelecanus. Sula is intermediate,
since in it the seventh vertebra does not bifurcate postaxially, as in Pelecanus, yet the
preaxial part of the eighth is more pressed back than that of the seventh of Pelecanus^
and less so than that of its eighth vertebra.

The postzygapophyses look more externad than either in Pelecanus or Sula.

The styloid processes are much larger than in the seventh vertebra.
. In the eighth vertebra the styloid process has come to start rather from the middle
than from the external border of the groove on each side of the preaxial part of the
median subcentral groove. It has thus become rather catapophysial than parapo-
physial, and the same change has taken place here as has been described as occurring
in the corresponding vertebra of Sula.

The NINTH VERTEBRA (Plate LVIL figs. 17-20) is, like the ninth of Sula, elongated;
but the neural spine is smaller, and the prsezygapophyses look more mesiad. It evidently
corresponds with the eighth of Pelecanus, though there is no haemal arch, it being the
first one in which the preaxial part of the vertebra is pressed in postaxiad, with a trans-
verse constriction behind the praezygapophyses. The three tubercles outside the pleura-
pophysial lamella are distinctly developed, with the parapophysial projection below them,
as in Sula. The rib-like styloid processes are larger and more slender than in the eighth
vertebra, and much more so than in the ninth vertebra of Sula.

The TENTH VERTEBRA (Plate LVIL figs. 21-23) is, like the tenth of Sula, somewhat
drawn out ; but there is no haemal arch 1. In this and the preceding vertebra in Sula
the rib-like styloid process is antero-posteriorly grooved on its ventral surface — not so in
Phalacrocorax.

The ELEVENTH, TWELFTH (Plate LVIL fig. 24), and THIRTEENTH VERTEBRAE are, like
those of Sula, somewhat lengthened ; but in none is the hremal arch completely closed
ventrally. In the eleventh vertebra the postzygapophyses fail, for the first time, to attain
the postaxial limit of the centrum. In the last of these vertebra? the styloid rib-like
process reverts to its condition in the seventh vertebra, inasmuch as it here again extends
from the external margin of the lateral antero-posterior subcentral groove; i. e. it is again
rather parapophysial ; and the parapophysial projection, distinct in the ninth, tenth,
and eleventh vertebras, has here coalesced with the root of the rib-like process.

The FOURTEENTH VERTEBRA is substantially like that of Sula's fourteenth vertebra ;
but when viewed dorsally it is more constricted laterally, the prsezygapophyses look more
directly mediad ; the neural spine is more developed, ending in a sharp dorsad and
postaxiad process at its postaxial end. Viewed laterally, a large, median, laterally
compressed bypapophysis here appears suddenly for the first time, largely perforated

do so already in the sixth vertebra. In all three genera the processes begin again to project postaxiad of the
articular surface in the fifteenth or sixteenth vertebra.

1 The styloid ribs are excessively long in this and adjacent vertebrae of 1182 c in the College of Surgeons.

VOL. x. — PART vii. No. 5. — August 1st, 1878. 3 B

348 PEOF. ST. GEOEGE MIVAET ON THE

towards its postaxial margin. The metapophysial ridge is more developed ; and beneath
it are two very strong antero-posterior ridges (separated by a marked concavity), outside
the pleurapophysial lamella, from the ventral postaxial angle of which a rib-like styloid
process extends postaxiad.

At the preaxial end of the ventral surface of the centrum there is a sharply limited
fossa just preaxial to the hypapophysis1.

In the next vertebra the postzygapophysial process again begins to extend postaxiad
beyond the articular surface — as also in Sula and Pelecanus.

The FIFTEENTH VEETEBEA (Plate LVII. figs. 25-29) is like the fourteenth, and the
fourteenth of Sula, except that the neural spine is a little more developed ; the styloid
processes are rudimentary. It differs from the fifteenth of Sula in that the neural spine
is so much less developed. The pleurapophysial lamella is wide, as in the fourteenth ver-
tebra of Sula. A lateral foramen sometimes leads into the substance of the bone ; only,
instead of being situated close to the ventral surface, it is just below the interzygapo-
physial ridge. The hypapophysis is largely developed.

The SIXTEENTH VRETEBRA closely resembles the fifteenth of Sula, only that there is a
large, long hypapophysis, instead of a slight median ridge. There is a lateral foramen,
as in the fifteenth vertebra.

The SEVENTEENTH VERTEBRA is very like the sixteenth of Sula, only that there is a
large hypapophysial process, that the lateral canal is still completed below by an osseous
bridge, that the hyperapophyses are more developed, and the postaxial articular surface
of the centrum more developed transversely.

The EIGHTEENTH VERTEBRA (Plate LVII. fig. 30) is like the seventeenth of /Sula, but
that there is a larger hypapophysial process, that the lateral canal is still completed
below by a delicate osseous bridge, that the hyperapophyses are more developed, that
the postaxial articular surface of the centrum is more extended transversely in propor-
tion to its dorso-ventral development, and that the postero-external angle of the centrum,
in its ventral aspect, is drawn out into a triangular process of bone, which extends
slightly ventrad. From the outer end of the ventral surface of the transverse process
a pointed process, more or less developed, extends externad and ventrad ; and sometimes3
there may be, instead of a ventral bony bridge, a long, delicate, movable rib, with
both capitulum and tuberculum.

CERVICO-DORSAL VERTEBRAE.

There are generally but two of these (Plate LVII. figs. 31-36).

The NINETEENTH VERTEBRA is like the eighteenth of Sula, except that there is a long
and narrow hypapophysis, that the prsezygapophyses are smaller and produced relatively
more dorsad, and that the hyperapophyses are drawn out into long, bony, very slender

1 This does not occur till the fifteenth vertebra in 1182 c.
As in No. 76. 9. 26. 16 of British Museum.

AXIAL SKELETON OF THE PELECANID.E. 349

processes. The centrum is more laterally compressed than in Sula ; and there is a ridge
running antero-posteriorly on each side of the origin of the hypapophysis.

The TWENTIETH VERTEBEA is like the nineteenth of Sula, except that it has become
distinctly opisthoccelous (fig. 36), and that there is a long hypapophysis, that it is
still more compressed laterally, and has no marked antero-posterior ridge on each
side of it.

THE DORSAL VERTEBRAE.

These are much more compressed laterally than in Sula or Pclecanus.

The TWENTY-FIRST VERTEBRA (Plate LVII. figs. 37-39) has on its preaxial articular
surface a median tract (of about one third of its total width) convex in both directions,
and external to this, on each side, a tract which is concave from within outwards. Its
postaxial articular surface is more simply and entirely concave than is that of the
twentieth vertebra. There is a long hypapophysis, a little antero-posteriorly expanded
at its distal end. This and the succeeding dorsal vertebra have styliform ossifications
of tendon attached to the dorsum of the distal part of their transverse processes ; and
the neural spines are, to a less extent, similarly affected.

The TWENTY-SECOND and TWENTY-THIRD VERTEBRAE (Plate LVII. figs. 40-45) are simply
convex in front, centrally, and concave behind, each surface being much more dorso-
ventrally than transversely extended. The hypapophyses are nearly as in the twenty-
first vertebra.

The TWENTY-FOURTH VERTEBRA is the first to form part of the ankylosed pelvic mass ;
its preaxial surface is like that of its serial predecessor, except that it is somewhat
broader. The hypapophysis is slightly smaller (Plate LX. fig. 6, xxiv).

The TWENTY-FIFTH VERTEBRA is like its predecessor, but is smaller, more compressed,
and with a smaller hypapophysis. Generally the hypapophysis of this vertebra is (like
its three serial predecessors) distinctly flattened ventrally ; but it may not be so, and
may be quite rudimentary, as in the smaller skeletons in the British Museum, 68. 8.
16. 3. 56. 11. 14. 17 and 66. 3. 28. 17.

THE LUMBAR VERTEBRAE.

The TWENTY-SIXTH, TWENTY-SEVENTH, TWENTY-EIGHTH, TWENTY-NINTH, and THIRTIETH
VERTEBRAE resemble the lumbar vertebra of Sula, save in their greater number, the
presence of distinct hypapophyses to the first two, and the presence, generally, of three
presacral foramina on each side, produced by the distal junction of the parapophyses of
the four most postaxial lumbar vertebrae; there may, however, be as many as four
presacral foramina.

THE LUMBO-SACRAL VERTEBRAE.

From the intervertebral foramina there appear generally to be two lumbo-sacral

3fi2

S50 PEOF. ST. GEOEGE MIVAET ON THE

vertebrae — the THIKTY-FIRST and the THIRTY-SECOND,— although there may be as many as
three, or sometimes, apparently, but one.

THE SACEAL VERTEBRAE.

Of these there are at most but two, the THIRTY-THIRD and the THIRTY-FOURTH.

They differ from those of Sula in that, though the first has, like that of Sula, its
parapophysis detached from more dorsal ossification, that of the second (thirty-fourth
vertebra) often, if not generally, has its parapophysis continuously ossified with what is
dorsal to it, so as to constitute a little subvertical sheet of bone formed of parapophysis
and diapophysis conjoined (Plate LXI. fig, 1),

THE SACRO-CAUDAL VERTEBRAE.

These are seven or eight in number, from the THIRTY-FIFTH to the FORTY-SECOND
inclusive (Plate LXI. fig. 1). The condition of their parapophyses is generally inter-
mediate between Pelecanus and Sula, not even arising so ventrally as in the former and
not generally bent up dorsad so suddenly as in the latter, except in the thirty-fifth
vertebra, where they are quite so. Their general appearance, however, is more like
that presented by Sula than by Pelecanus. They are subequally developed as to the
transverse extent of their parapophyses ; and the last two considerably increase in antero-
posterior extent.

THE CAUDAL VERTEBRAE.

These vertebrae, from five to eight in number, rapidly increase in transverse diameter ;
and the second may have a hypapophysis, while the third has generally a distinctly
bifurcating hypapophysis, which becomes much larger in the succeeding two vertebrae.
Each vertebra has a pair of small zygapophyses (Plate LXI. fig. 5).

The PYGOSTYLE is very different in shape from that of either Pelecanus or Sula. It
is much shorter and sharply bent up dorsad upon the free caudals ; its ventro-postaxial
margin is gently concave, and its dorso-preaxial one gently convexo-concave postaxiad —
both as in Sula ; but it is a very much shorter bone, and may even approximate in
outline to an equilateral triangle, though it may be a triangle of which the breadth is
but half the length.

THE PELVIS.

This includes as many as from fifteen to seventeen vertebrae. It more nearly resembles
that of Sula than that of Pelecanus, but is more elongated antero-posteriorly.

Viewed laterally it agrees with Sula, and differs from Pelecanus in the points men-
tioned in describing Sula, except that the sacro-iliac ankylosis is longer, and the ilio-
caudal spine is a very long pointed process (Plate LX. fig. 6).

Compared with Sula, the sciatic foramen is more pointed postaxially, and, as in

AXIAL SKELETON OF THE PELECANIDJL 351

Pelecanus, narrower (dorsoventrally) behind than in front, instead of the reverse, as
generally in Sula.

The pubis is shorter and curved more sharply ventrad at its postaxial end.

Viewed dorsaUy, the antero-posterior excess in length of the postacetabular part is
greater ; and the ilia expand at their preaxial ends in a wing-like manner, the external
margin of the ilium behind the expansion being much more concave, while a strong but
narrow ridge divides the two iliac fossa?. The sacro-sciatic foramina are about as visible as
in Sula ; but the postaxial margin of the pelvis is very different, the ilio-caudal spines (Ic)
projecting postaxiad generally more, often considerably more, than the ischium, and
nearly as much as the pubis. The angle formed by this spine with the adjacent margin
of the ischium is acute. The transverse diameter of the pelvis at the antitrochanteric
processes is even exceeded by that of the preaxial ends of the ilia. The postaxial half
of the external margin of the ischium is convex instead of being concave as it is in
Sula. The defects of pelvic ossification generally extend preaxiad to quite between
the acetabula (Plate LXI. fig. 2).

Ventrally examined (Plate LXI. fig. 1), the pelvis of course repeats the characters
which can be seen thus as well as by the dorsal view ; and, in addition, the ventral surface
of the most preaxial part of each ilium extends more out laterally beyond the ends of
the transverse processes than even in Sula. There are generally three small presacral
foramina on each side — the first the largest, the second the longest, the third the smallest
and round.

The lateral acetabular fossae (f) are smaller than even in Sula, and rhomboidal in out-
line. The parapophysis of the second sacral vertebra, ankylosing, as it often if not gener-
ally does, with the ilium above, forms a complete postaxial boundary to the lateral fossa
in front of it — that fossa extending above (dorsal to) the parapophysis of the first sacral
vertebra — and at the same time to the posterior iliac fossa behind it. Thus there is no
trace whatever of the large opening which extends in Pelecanus above the expanded
ends of the sacral parapophyses. There is no supraacetabular fossa.

The parapophyses of the last lumbar vertebra abut directly against the preaxial
margin of the acetabulum, instead of some distance preaxial to it as in Sula and
Pelecanus. The antero-posterior ridge inside the ischium, bounding its groove (g)
mesially, is more sharply marked than in Sula or Pelecanus ; and the extent of ischium
ventrad of the more postaxial part of that ridge is very much greater.

THE ILIUM.

This bone extends over the twenty-fifth (or last dorsal) vertebra and part of the one
preceding.

Seen laterally, the dorsal margin of the preacetabular part is slightly more convex,
and its ventral margin is more convex preaxially and concave postaxially ; there is a
slight indication of an ilio-pectineal process (Ip).

352 PEOF. ST. GEOEGE MIYAET ON THE

The ilio-caudal process (Ic) is pointed and very long, extending directly postaxiad and
beyond the postaxial end of the ischium. The sacro-sciatic foramen is longer and more
pointed postaxially; at the anterior part of its dorsal margin a very distinct post-
trochanteric process (st) is developed postaxial to the antitrochanteric process, removed
from it by a distance rather less than the diameter of the acetabulum1.

THE PUBIS
is more sharply bent down ventrad towards its distal end.

THE ISCHIUM

has its external smooth antero-posterior groove hardly, if at all, deeper than in Pelecanus.
The process (ps) in which it ends preaxially, is rather more marked than in Sula.

The internal (ventral) groove (g) is much as in Pelecanus, but more expanded towards
its postaxial end. The part postaxial to the sciatic foramen is more expanded than in
Sula, and is more as in Pelecanus.

The ventral surface of the ischium behind the sciatic foramen (and mediad of the
very strong ridge bounding medianly the antero-posterior ventral groove) is smooth ; it
is less concave than in Pelecanus, but more extensive than in Sula.

THE RIBS.

There are seven vertebral and five sternal. There may be eight vertebral ribs when
the eighteenth cervical vertebra bears one.

THE VERTEBRAL RIBS.

The first two have no sternal ribs ; all, or all but the last, have uncinate processes ;
these processes are more developed than in Sula.

The last ends proximally in a filamentary process, and then ankyloses with the pelvis.
The first rib joining the sternum is the broadest.
None are much expanded distally.

THE STERNAL RIBS.

These increase in size postaxiad (Plate LXI. fig. 3).

The last is not double generally, but may be so.

The fourth is twice as long as the first ; the fifth is longer.

None are very much expanded distally.

The fourth and fifth may unite before joining the sternum.

1 In 1182 c this process becomes an antero-posteriorly extended ridge.

AXIAL SKELETON OF THE PELECANHLE. 353

THE STERNUM.

The sternum (Plate LXI. figs. 3 & 4) is shorter and broader than in Sula, with the
proportions of Pelecanus, but less contracting, laterally, postaxiad. The clavicles are
not ankylosed to it.

The keel extends much relatively as in Pelecanus ; and the coracoid grooves fornTan
angle much as in that genus, and therefore a less acute one than in Sula. The median
xiphoid (mx) agrees with that of Sula in not extending postaxiad as much as the lateral
xiphoids do, thus differing from Pelecanus. These lateral xiphoids are obtusely pointed,
not spatulate (Ix),

The pleurosteon may have only four elongated articular processes in Ph. carbo, but
generally has five such in other species.

The concavity and convexity of the sternal osseous sheet is much as in Pelecanus,
and not flattened as in Sula, though the middle part of the postaxial portion of its
ventral surface is not so convex transversely as in Pelecanus.

The articular surface behind the coracoid groove is not so broad relatively as either
in Sula or Pelecanus.

A process may spring preaxiad from dorsal end of concave dorso-preaxial margin
of keel.

PLOTUS.

The anatomy of Plotus has been recently described by Professor Garrod, in a very
interesting and instructive memoir in the ' Proceedings ' of the Zoological Society
for 1876, p. 335, pis. 26 to 28. Its skeleton was noticed by Mr. Eyton in his
' Osteologia Avium,' p. 218 ; and by W. Donitz (as concerns the cervical vertebrae)
in the 'Archiv fur Anatomic und Physiologic,' 1873, p. 357, plate ix. A.

It has been described with much detail by Brandt, who has given figures of the entire
skeleton, the pelvis, and sternum, in the ' Memoires de 1' Academic des Sciences de
St.-Petersbourg,' 6me serie, Sciences Mathematiques, Physiques et Naturelles, tome v.
seconde partie, Sciences Naturelles, tome iii. 1840, p. 132, pi. 4.

I have myself examined a skeleton of Plotus anhinga in my own collection, and also
another and a skeleton of P. novce-hollandice, both in the Museum of the Royal College
of Surgeons.

The total number of vertebrae seems to be 45 or 46, without counting the pygostyle.

These vertebrae are subdivisible as follows — 18 cervical vertebrae (in the specimen
of P. novce-hollandice there are only 17; but this is probably owing to one vertebra,
the 15th, having been lost), 2 cervico-dorsal, 5 or 6 dorsal (in one specimen of P.
anhinga there are but 5), 4 lumbar, 2 or 3 lumbo-sacral, 2 sacral, 4 sacro-caudal, and
7 or 8 caudal vertebrae.

354 PKOF. ST. GEOKGE MIVART ON THE

The whole axial skeleton, when compared with those of the genera previously
described, presents a remarkable contrast in the great prolongation of its more preaxial
part, while its pelvic part is somewhat shortened, notably as compared with the same
part in Phalacrocorax, with which genus, nevertheless, the axial skeleton of Plotus, on
the whole, most agrees. Thus it agrees with Phalacrocorax in the strongly opistho-
crelous condition of the trunk-vertebrae, and in their median, subcentral, laterally
compressed prominences, though (as we shall see) they have not the same nature as
the analogous parts in Phalacrocorax.

In the elongation of the anterior cervical vertebrae Plotus is more nearly approached
by Pelecanus; but these bones are not nearly so long and slender relatively in the
last-named genus, nor are they so smooth dorsally and with such rudimentary neural
spines as in Plotus.

THK CERVICAL VERTEBRA.

These vertebrae, from their length and the little relative prominence of their pro-
cesses, resemble (as has just been said) the cervical vertebrae of Pelecanus ; but in
number and essential details of structure they rather resemble those of Phalacrocorax,
presenting at the same time a number of special peculiarities.

As has been pointed out by Professor Garrod1 the vertebrae are mostly long to the
eighth, which is extraordinarily elongated (as also the third, fourth, and fifth), and
forms an angle open ventrad with its predecessor, and an angle open dorsad with its
successor.

The styloid rib-like processes are little developed in the first seven vertebrae, but in
the eighth and succeeding ones much so, above all in the eighth, where their length is
such as to extend peculiarly, ventrad, and postaxiad of the preaxial end of the haemal
arch of the vertebra next behind, i. e. the ninth.

The ATLAS is much like that of Phalacrocorax ; but there are no lateral hypapophysial
processes, and the hyperapophyses project less postaxiad, scarcely more so than in
Pelecanus. The odontoid foramen is smaller, relatively as well as absolutely, than
in any of the three preceding genera (Plate LVIII. figs. 1 & 2).

The AXIS differs remarkably from that of all the three preceding genera in its great
relative antero-posterior length and little dorso-ventral extent, due principally to the
small development of the hypapophysial element; it is also relatively less broad
transversely (Plate LVIII. figs. 3 & 4).

The odontoid process is very small. There is a distinct, elongated, though low, neural
spine. The articular surfaces of the postzygapophyses project as much postaxiad as do
the zygapophyses themselves. There is an interzygapophysial ridge. The postaxial
articular surface of centrum looks dorsad as well as postaxiad ; and its ventral margin is
strongly concave. The ventral surface of the centrum presents three antero-posterior

1 Proc. Zool. Soc. 187C, p. 33&-T.

AXIAL SKELETON OF THE PELECANID^. 355

ridges : — one median, hypapophysial ridge, extending to about the postaxial end of the
preaxial two thirds of that surface ; and two lateral, parapophysial ridges (fig. 4, hi/'), ex-
tending and diverging postaxiad till each ends in a process on the ventral side of the ex-
ternal ventral angle of the postaxial articular surface of the centrum. There thus comes
to be a small fossa beneath the postaxial third of the centrum. The postzygapophyses
project postaxiad much beyond the postaxial end of the centrum.

In Plotus novce-hollandice the hypapophysial ridge is more developed, and extends
almost to the postaxial end of the centrum, while the lateral ridges are like delicate
rib-like styloid processes, there being a defect of ossification in each.

The THIRD VERTEBRA (Plate LV1II. figs. 5 & 6) is almost half as long again as is the
second. It is, like the third of Pelecanus, greatly drawn out, and with the neural spine
and hypapophysis all but entirely suppressed. There is no interzygapophysial ridge.
The postaxial half of the median hypapophysial ridge is replaced by a narrow, shallow
groove, the margins of which must be reckoned as catapophysial, while a parapophysial
ridge runs along each side of the ventral surface of the centrum, and ends postaxially
in a rib-like styloid process, which begins to project freely ventrad as well as postaxiad
at about the preaxial end of the postaxial fourth of the centrum, almost extending
postaxiad as far as the preaxial end of the postaxial articular surface of the centrum.
A small tubercle (which may be either the representative of a tubercular process or
a metapophysis) projects just behind the praezygapophysis and above the tubercle at
the preaxial end of the parapophysial ridge. From the former tubercle a slight ridge
descends obliquely to join the parapophysial ridge at about its middle.

The FOURTH VERTEBRA (Plate LVIII. figs. 7 & 8) is about equal in length to the third,
and is quite like it, except that the median subcentral groove is a little deeper and
extends along the whole length of the centrum. At the postaxial end of the preaxial
third of each lateral margin of the subcentral groove a small catapophysial process
projects downwards, while between such lateral margin and the more externally situated
preaxial part of the parapophysial ridge (p1 ) is an antero-posterior groove (fig. 8, g) which
corresponds with the similarly situated groove beneath the fifth and sixth vertebrae of
Pelecanus. The styloid process begins to start out at a point slightly more preaxial.

The FIFTH VERTEBRA (Plate LVIII. figs. 9-11) is like the fourth, except that it is
very slightly shorter, and that the styloid processes are again a little more preaxial,
and the postzygapophyses extend postaxiad a little beyond their articular surfaces.

The SIXTH VERTEBRA (Plate LVIII. figs. 12-14) is decidedly shorter, not greatly ex-
ceeding the axis in length ; and here, while the catapophysial margins of the sub-
central groove are more raised, the styloid processes are shorter and more preaxial,
a wide, subequal, antero-posterior groove extending between the catapophysial and para-
pophysial ridges. Here a neural spine begins to reappear1, the postzygapophyses are

* In Plotus novw-liollandice it begins in the fifth vertebra.

VOL. x. — PART vii. No. 6. — August 1st, 1878. 3c

356 PEOF. ST. GEOEGE MIVAET ON THE

rather more distinct, though still small, and the hyperapophyses are decidedly more
marked than in the fifth vertebra.

The SEVENTH VERTEBRA (Plate LVIII. figs. 15-17) is, again, a trifle shorter, but little
exceeding the axis in length. Compared with the last vertebra, the styloid processes
are yet shorter, and (as in that vertebra) project but very little from the side of the
centrum, and are separated from the adjacent catapophysial ridges by a wider, though
somewhat shorter, antero-posterior groove. The neural spine is higher ; the hypera-
pophyses and the metapophyses are stronger; and the postzygapophyses almost look
a little preaxiad as well as ventrad, and a little externad. As in the sixth vertebra,
so here the plane of the styloid processes was more dorsad than that of the hinder
part of the centrum's ventral surface. The centrum being a little bent on itself
from before backwards, its ventral margin (when viewed laterally) is a little concave
ventrad.

A narrow fissure runs preaxiad from the middle of the postaxial margin of the neural
arch ; and on each side of this, between it and the adjacent hyperapophysis, is an antero-
posteriorly directed groove.

The EIGHTH VERTEBRA (Plate LVIII. figs. 18-22) agrees with that of Phalacrocorax,
and differs from that of Sula and Pelecanus, in that it articulates with its predecessor
at an angle open ventrad, as in the more anterior vertebra, while with its successor it
articulates at an angle open distad ; but it presents characters altogether peculiar. As
in the seventh of Pelecanus and the eighth of Phalacrocorax, the neural spine is more
preaxially situated than in its serial predecessor (projecting from about the antero-
posterior middle of the neural arch) ; and, as in the seventh of Sula and eighth of Pha-
lacrocorax, the styloid processes diverge ventrad from the centrum as they proceed post-
axiad. Unlike, however, all the three before-described genera, these styloid processes
are enormously long, extending more postaxiad than any other part of the vertebra,
and reposing upon the sides of the hsemal arch of the ninth vertebra. The postzygapo-
physes also do not project postaxiad as much as does the centrum ; and their articular
surfaces (which are elongated) look more directly externad than even in Phalacrocorax —
though, as in that genus, they differ from those of the sixth vertebra by quite extending
to the postaxial ends of those processes. The preaxial part of the neural arch is not
pressed back postaxiad ; but the postaxial part of that arch rises dorsad — instead of being
curved over ventrad very markedly, as it is in the seventh vertebra; also the postaxial
margin of the neural arch, seen dorsally, is more concave than in the preceding vertebra,
and without any narrow median cleft. The prsezygapophyses still extend decidedly pre-
axiad of the centrum. The long styloid processes cannot well be determined as more
catapophysial or parapophysial ; for they arise from the same part as the parapophysis
arises from in the seventh vertebra, while they are continuous dorsally with the preaxial
part of the marked catapophysial ridges, which bound laterally the median subcentral
groove ; these latter ridges are externally concave dorso-ventrally, this concavity being

AXIAL SKELETON OF THE PELECANIDJ?. 357

bounded above by an antero-postevior ridge. This lateral groove ends preaxially in a
small, but deep, fossa enclosed externally by the external root of the styloid process.

The NINTH VERTEBRA (Plate LVIII. figs. 23-27) is very much shorter than the eighth,
being scarcely, if at all, longer than the axis. It agrees with the ninth vertebra of Pha-
lacrocorax in being the one to present the marked contrast, to be the first pressed back
postaxiad at its preaxial part, and to be articulated with its two adjacent vertebrae at angles
both open dorsad. It thus agrees with the eighth vertebra of Pelecanus and of Sula.

Here in Plotus is a haemal arch for the first time developed, and developed suddenly,
and at once in its completeness. It extends much preaxiad of the centrum, while
neither pair of zygapophyses may extend quite as far, pre- and postaxially respectively,
as does the centrum ; the prgezygapophyses never do so.

The neural arch is much cut away preaxially ; and the preaxial articular surface of
the centrum looks mainly dorsad. The praezygapophysial surfaces look mediad and
very slightly postaxiad. These processes are produced dorsad and postaxiad like those
of the tenth vertebra of Phalacrocorax and the ninth of Pelecanus.

The neural spine is very small, and close to the concave postaxial margin of the
neural arch. Hyperapophyses are more developed, and have retreated preaxiad from
the ends of the postzygapophyses. In Plotus anhinga, immediately in front of each
hyperapophysis is another small process (fig. 24, 7«j/), of smaller size ; to this one end
of a strong tendinous arch is attached, the other end of which is implanted in the dorso-
postaxial angle of the prsezygapophysis of the same side. In P. novce-hollandice this tendon
is ossified, forming a bony arch. These bony arches arise from a point on the dorsum
of the neural arch which seems to represent spine and hyperapophysis ; and no neural
spine is developed. Preaxially each arch goes to the metapophysial prominence outside
the prsezygapophysis. We here find, for the first time in the series, a conspicuous
lateral canal, which is short, and opens preaxially dorsally to the centrum. Oblique
metapophysial ridges project from the outer surface of the praezygapophyses. The
haemal arch is notched medianly preaxially, with a small, ventrad, median prominence
behind it. The styloid processes reach almost to the postaxial end of the centrum. It
is hard to say whether they are catapophysial or parapophysial in nature ; but marked
catapophysial ridges bound on each side the median, subcentral canal.

The TENTH VERTEBRA (Plate LVIII. figs. 28-30) is slightly longer than the ninth, and
is like it except that the postaxial margin of the neural arch is a little more concave,
that there is no neural spine, that the hyperapophyses are larger, the styloid processes
are shorter, the outside of the prtezygapophysis is more rugged, and the postzygapophyses
begin always to project again a little postaxiad of the postaxial end of the centrum.
The haemal arch has a strong median ridge running antero-posteriorly on its ventral
surface ; and the catapophysial ridges (bounding laterally the median, subcentral groove)
are, as usual in Plotus, strongly marked.

The ELEVENTH VERTEBRA (Plate LVIII. figs. 31 & 32) is of about the same length as

3c2

358 PROF. ST. GEOKGE MIVAET ON THE

the tenth, and is like it except that the prsezygapophyses again extend a little less
dorsad (as they have begun to do in the tenth vertebra), while from behind the summit
of each a sharp, pointed, metapophysial process extends dorsad, postaxiad, and externad.
The hyperapophyses are more developed (especially elongated) ; and there is no longer
any trace of a neural spine.

The median ventral ridge of the hsemal arch has here become a very sharp, plate-like,
prominent process.

The TWELFTH VERTEBRA (Plate LVITI. fig. 33) is of about the same length as the
eleventh, and is like it except that the hind ventral process of the ha?mal arch is more
prominent, the hyperapophyses are larger, and the postzygapophyses here again begin
to fail to extend postaxiad so far as does the centrum.

The THIRTEENTH VERTEBRA is slightly shorter than the twelfth, and repeats its cha-
racters a little more strongly marked, except that the haemal arch projects preaxially
to the centrum in a somewhat less degree. The hyperapophyses project sharply and
strongly dorsad and preaxiad.

The FOURTEENTH VERTEBRA is again a little shorter. The postzygapophyses begin to
project postaxiad slightly beyond their articular surfaces. It has no longer the pointed
process just postaxial to the prszygapophysis, or only a rudiment of such. Long cata-
pophysial ridges limit laterally the subcentral groove, though they altogether disappear
in the next vertebra.

The FIFTEENTH VERTEBRA (Plate LVIII. figs. 34-36) is of about the same length as the
fourteenth. It is like it ; but the rib-like styloid processes are stouter, more externally
placed, and are more decidedly parapophysial. The haemal canal has ceased to exist ;
and in its place is a long plate-like hypapophysis, which may be perforated.

The hyperapophyses converge preaxiad, and may meet to indicate a rudiment of a
neural spine1. The postzygapophyses begin again to project postaxiad a little beyond
the centrum.

The SIXTEENTH VERTEBRA is of about the same size as its serial predecessor, but differs
from it markedly in the sudden development of a long neural spine, curved postaxiad
at its tip. There is also a still larger plate-like hypapophysis, projecting postaxiad of
the centrum, and considerably larger than in the corresponding vertebra of Phala
crocorax. The prsezygapophyses begin again to extend preaxiad as much as does the
centrum. The rib-like styloid processes are shorter and stouter, and notched preaxially,
•while the lamella, enclosing dorsally the lateral vertebral canal, extends more postaxiad,
and is therefore longer antero-posteriorly than in the fifteenth vertebra.

The SEVENTEENTH VERTEBRA is of about the same size as the sixteenth ; but its neural

1 This vertebra, in the specimen of P. novce-hollandice in the Eoyal College of Surgeons' Museum, has a neural
spine more developed than has even the sixteenth vertebra of P. anhinga (in either specimen), -while its four-
teenth vertebra is quite like the fourteenth vertebra of P. anhinga. This, coupled with the defective number
of cervical vertebrae, makes it probable that the fifteenth is lost in this specimen.

AXIAL SKELETON OF THE PELECANUm 359

spine is much more extended antero-posteriorly, while its hypapophysis is much reduced
(not being larger than in Phalacrocorax}, but still extending postaxiad of the centrum.
The rib-like styloid process is still shorter and more notched preaxially; while the
lamella of bone enclosing dorsally the lateral vertebral canal is still more antero-
posteriorly extended, and also more outwardly. The ventral surface of the centrum is
flattened and much broader than in the sixteenth vertebra, while its external margins
are developed with lateral catapophysial ridges.

The EIGHTEENTH VEETEBEA (Plate LVIII. figs. 37 & 38) is like the seventeenth, except
that the neural spine is still more elongated antero-posteriorly, and dorsally flattened,
and the hypapophysis still more diminished, the centrum still more flattened beneath,
and its lateral ridges more developed. While the lamella enclosing the lateral canal
dorsally is as large, there is ventrally but a slender osseous bridge, the point of junc->
tion between which and the dorsal plate is the last rudiment of the styloid process.

THE CERVICO-PORSAL VERTEBRA,

In the NINETEENTH VERTEBRA, compared with the eighteenth, the transverse processes
extend less outwards and are inclined more postaxiad, and the lateral hypapophysial
ridges are more developed, while the root of the median hypapophysis extends further
preaxiad (Plate LVIII. figs. 39 & 40).

Compared with the nineteenth vertebra of Phalacrocorax, the transverse processes
extend much less outwards, the hypapophysis is much more postaxially situated, and
more inclined postaxiad distally. The centrum is very much less laterally compressed,
and very much more flattened ventrally, with lateral hypapophysial ridges much more
laterally extended ; these ridges cannot be parapophysial, because the heads of the ribs
articulate dorsally to them. The postaxial articular surface "of the centrum is much more
extended transversely; and the prgezygapophyses are relatively as well as absolutely
smaller.

In the TWENTIETH VERTEBRA (Plate LVIII. figs. 41-43), compared with the nineteenth,
the lateral hypapophysial plates (or catapophyses) are still more developed externad,
being conspicuous lamellar processes. They do not, however, now reach to the post-
axial end of the centrum, but only project out from its preaxial half. The median
hypapophysis is shorter dorso-ventrally ; but its root now extends to the preaxial margin
of the centrum. The postaxial surface of the centrum is less extended transversely.

Compared with the twentieth vertebra of Phalacrocorax, the median hypapophysis is
very small and postaxiad, while the centrum, instead of being greatly compressed
laterally, is extremely flattened ventrally with great lateral hypapophysial processes,
the distal ends of which begin to incline somewhat ventrad. The outline of the post-
axial articular surface of the centrum is more quadrangular, not rounded, and is less
concave. Its sides are concave from the outward extension of the postaxial ends of the
lateral hypapophysial ridges.

360 PEOF. ST. GEOEGE MIVAET ON THE

This vertebra, however, agrees with the corresponding one of Phalacrocorax in that
in it the praezygapophyses again, for the first time, do not extend so far preaxiad as the
preaxial end of the centrum.

THE DORSAL VERTEBRA.

These vertebrae (Plate LVIII. fig. 46) agree with the dorsal vertebrae of Phalacro~
corax in being laterally compressed and strongly hypapophysial, or rather catapophysial,
since the hypapophyses are differently -formed, i.e. from rnedianly coalescing lateral
hypapophyses.

The TWENTY-FIRST VERTEBRA differs from the twentieth in that the median hypapo-
physis has all but disappeared (being only a rudimentai'y median ridge), while the lateral
hypapophyses, though reaching to the postaxial end of the centrum on each side as
lateral ridges, yet have their long, outstanding, distally depressed processes confined to
a rather smaller extent of the preaxial part of the centrum. The preaxial surface of the
centrum is less concave, its postaxial surface more so.

Compared with the same vertebra in Phalacrocorax, the preaxial articular surface of
the centrum is more quadrangular, the centrum is less compressed laterally and more
ventrally flattened. The median hypapophysis has disappeared ; but there are con-
spicuous lateral hypapophyses. The transverse processes project less outward, but are
more antero-posteriorly developed. The fossa beneath the root of the transverse process,
which is so conspicuous in Phalacrocorax, is here wanting.

The TWENTY-SECOND VERTEBRA (Plate LVIII. fig. 44) differs from the twenty-first mainly
on its ventral aspect. Its centrum is much more laterally compressed ; and the median
hypapophysis has quite disappeared; the lateral hypapophyses (already in the last
vertebra slightly more ventrad than in the twentieth vertebra) have coalesced medianly
under the postaxial half or more of the centrum, while their preaxial ends project out-
wards, downwards, and backwards, but are much smaller than in the twenty-first ver-
tebra. This vertebra is also more decidedly opisthoccelous than its serial predecessor.

Compared with the same vertebra of Phalacrocorax, the centrum is slightly more
preaxially convex (especially dorso-ventrally) and postaxially concave. The transverse
processes differ much as do those of the twenty-first vertebra. There is no true median
hypapophysis ; but, as has been said, the lateral hypapophyses bend down and unite to
form postaxially a coalesced, vertical, hypapophysial lamellar process.

The TWENTY-THIRD VERTEBRA (Plate LVIII. fig. 45 & Plate LXI. figs. 6, 7, 8, xxui) differs
from its predecessor mainly in that its centrum is still more laterally compressed, and
that the lateral hypapophyses have still more completely united medianly, there being
now only two small diverging processes at the preaxial end of a median, vertical, sub-
Central plate, the ventral outline of which is concave antero-posteriorly.

Compared with the same vertebra of Phalacrocorax, its centrum is even more laterally
compressed ; the hypapophysis is also shorter, and, as has been said, its distal preaxial
end bifurcates.

AXIAL SKELETON OF THE PELECANID^E. 361

The TWENTY-FOURTH VERTEBRA (Plate LXI. figs. 6, 7, 8, xxiv) greatly resembles its prede-
cessor ; but its hypapophysial plate scarcely bifurcates at all preaxially, while its ventral
margin is more concave antero-posteriorly. It is extremely like the same vertebra of
Phalacrocorax ; and the non-identity in nature of the hypapophysial processes of the two
genera would never be suspected were these two vertebrae alone compared without their
serial predecessors.

The TWENTY-FIFTH and TWENTY-SIXTH vertebrae are successively smaller and smaller,
and have the hypapophysial plate successively simpler and less developed.

They differ from the corresponding vertebrae of Phalacrocorax in having the hypapo-
physial process smaller.

The twenty-sixth vertebra may be devoid of any vertebral rib l ; but a sternal rib
always corresponds with it.

THE LUMBAR VERTEBRA.

These appear to be constantly four in number, with short parapophysial processes
ankylosing distally with the ilium, and thus intercepting between them three anterior
pelvic foramina on each side. These vertebrae (the TWENTY-SEVENTH, TWENTY-EIGHTH,
TWENTY-NINTH, and THIRTIETH) have their centra continuously ossified together and with
the last two dorsal, and with the lumbo-sacral, sacral, and sacro-caudal vertebrae. The
centra of the lumbar vertebrae gradually broaden postaxially ; and though the first has
a hypapophysial prominence little less than that of the last dorsal, yet the two most
postaxial have a median, ventral, antero-posterior groove. The parapophyses become
successively shorter postaxiad. They thus present a general similarity to the lumbar
vertebrae of Phalacrocorax (Plate LXI. fig. 8).

THE LUMBO-SACRAL VERTEBRAE.

Of these there may be two or three ; they are the THIRTY-FIRST and THIRTY-SECOND
VERTEBRAE of my specimen, which I have taken as my type. They are generally similar
to those of Phalacrocorax.

THE SACRAL VERTEBRA.

There seem to be constantly two of these vertebrae (the THIRTY-THIRD and THIRTY-
FOURTH) ; but they are less differentiated from the sacro-caudals than is the case in
Phalacrocorax. Yet the parapophysis does descend decidedly more ventrad than does
the transverse process in the anterior sacro-caudal vertebrae. Moreover, as in Phala-
crocorax, the transverse process of at least the more posterior sacral vertebra forms
one continuous whole — a somewhat, though but slightly, vertically extended plate.
On the other hand, in the first sacral, the parapophyses may be more or less separated
from the diapophysis.

1 As in the specimen of Plotus anhinga in the College of Surgeons' Museum.

362 PKOF. ST. GEOEGE MIVART ON THE

THE SACKO-CAUDAL VERTEBRAE.

These are four in number; and the first two seem to be devoid of parapophyses, thus
making a more or less marked distinction between them and the sacral vertebrae.
Compared with those of Phalacrocorax, they are less elongated antero-posteriorly, less
laterally compressed as to their centra, with shorter transverse processes.

THE CAUDAL VERTEBRA.

There are seven or eight caudal vertebrae, not counting the pygostyle. They resemble
those of Phalacrocorax ; but their parapophyses are not so much bent dorsad, and are
more dorsal at their origin in the preaxial vertebrae, descending to a slightly more
Ventral origin as we proceed postaxiad.

The PYGOSTTLE is even more compressed laterally than in Phalacrocorax, and thus
differs even more than does the latter from the pygostyle of Sula. On the other hand,
it is much more curved, like the claw of a large Felis (Plate LVIII. fig. 47).

THE PELVIS.

This bony mass includes fifteen or sixteen vertebras, extending preaxiad over the
twenty-fifth (and sometimes even over part of the twenty-fourth) vertebra. It is very
much less elongated, relatively as well as absolutely, than in Phalacrocorax, and is
shorter, relatively, than even in Sula, so as to be a diminutive representation of that of
Pelecanus as to its general proportions, though in certain details it most nearly
approaches the pelvis of Phalacrocorax.

Viewed laterally^ the sacro-ischiatic foramen is seen to be less elongated than in Pha-
lacrocorax, and relatively broader dorso-ventrally. The union of the ilium and ischium
postaxial to it is short, shorter than in any of the before-described genera except in
Sula, while the ischium appears relatively deeper dorso-ventrally than in any. The
obturator-foramen is also smaller than in any of the other forms. The ilio-caudal
spine is long and pointed, as in Phalacrocorax (Plate LXI. figs. 6 & 7).

Viewed dorsally, the most striking features of the pelvis are the strong, curved
posterior iliac ridges which bound the sacral part laterally, which part is relatively
wider than in any preceding form save Pelecanus. These ridges absolutely distinguish
Plotus from all the other three genera (Plate LXI. figs. 6 & 7, Ir).

The antero-posterior length of the preacetabular part of the pelvis, as in Phalacro-
corax, is much less, compared with that of the postacetabular part, than in Pelecanus
or Sula ; and it agrees with that of Phalacrocorax in the wing-like lateral expansion of
its most preaxial part, and in the strong median ridge.

The ilio-caudal spine projects postfixially little, if at all, beyond the ischium ; and the
angle formed by it with the adjacent margin of the ischium is less acute.

The external margin of the postaxial half of the ischium is convex, as in Phalacro-
corax, not concave as in Sula.

AXIAL SKELETON OF THE PELECANIDJE. 363

Seen ventmlly, the ilia extend out as in Phalacrocorax ; and there are three presacral
foramina (a1, s2, s3) on each side. The lateral acetabular fossae (f) are notably different
from those of Phalacrocorax, owing to the encroachment inwards of the margins of
the acetabula, which almost reduce the preaxial part of each fossa to a narrow chink
(Plate LXI. fig. 8).

The ridge running postaxiad inside the ischium is less marked ; and there is no such
expanse of ilio-ischial surface dorsad of the postaxial part of such ridge as there is in
all the other three genera, above all in Pelecanus. On the other hand, the expanse of
the ischial surface ventrad of the postaxial part of such ridge is greater relatively than
even in Phalacrocorax, and therefore very much greater than in Sula or Pelecanus,
The pubes do not approximate ventrad as in Phalacrocorax or Sula.

THE ILIUM.

The process (st) described as existing in Phalacrocorax a little postaxial to the antitro-
chanteric process also exists here, but is nearer to the last-mentioned process. This
more postaxial process may be called the posttrochanteric process (Plate LXI. fig. 7).

Besides the points above noted, a ridge runs mediad and preaxiad from this post-
trochanteric process, and joins its fellow of the opposite side in the median antero-
posterior iliac ridge. Another ridge (Ir), very much stronger indeed, runs postaxiad from
the same process, forming the very prominent ridge of the postacetabular part of the
ilium, before noticed. This ridge causes each postacetabular ilium to present three
surfaces, one externad, one dorso-mediad, and the other ventro-mediad; and the pro-
minent ridge in question divides the externad from the dorso-mediad surface.

The ilio-pectineal prominence (lj)} is more sharp and prominent than in any other form
of Pelecanidae.

THE PUBIS.

This is but a very feeble ossification, and unites with the ischium for a longer space
than in the other Pelecanidee.

THE ISCHIUM.

This bone differs from that of the other Pelecanidae by its greater postaxial expansion
dorso-ventrally, presenting an extensive grooved surface (fig. 8, g) ventrally, as already
noted in describing the ventral aspect of the pelvis.

THE KIBS.
There are seven or eight vertebral and six sternal ribs.

THE VERTEBRAL HIBS.

The first two belong to the cervico-dorsal vertebra, and have no corresponding sternal ribs.

The second may or may not have an unciform process ; and the same is the case with
the seventh rib. The eighth, when present, has no such process.

The last rib has no tubercular process.
VOL. x. — PART vn. No. 7. — August 1st, 1878. SD

364 PKOF. ST. GEORGE MIVAET ON THE

THE STERNAL KIBS.

There are six of these on each side ; and the last does not join the sternum, but joins
the postaxial side of the penultimate sternal rib. The fourth is a little more than twice
as long as the first ; and the fifth is a little longer. The second, third, and fourth expand
distally to a moderate extent from within outwards (Plate LXI. fig. 9).

THE STERNUM.

This bone (Plate LXI. figs. 9 & 10) is quite like the corresponding bone of Phala-
crocorax, except that there are but four distinct articular surfaces on the pleurosteon,
while just postaxiad to the pleurosteon an angle (a) projects outwards, just internal to
which passes the fifth sternal rib. This projecting angle distinguishes the sternum of
Plotus from that of any other of the Pelecanidae. The keel approaches the postaxial
margin of the sternum more nearly than in Phalacrocorax or than in any other of
the Pelecanidee ; and the lateral xiphoids are narrower and more pointed.

Besides the four genera described, the two genera Fregata and Phaeton are usually
classed with them to constitute the group of the Steganopodes. But, from the point of
view here adopted (that of the postcranial part of the axial skeleton only), I have found
it impossible to detect characters which seem to me good and sufficient to unite such
Steganopodal group together and at the same time divide them off from other forms.
The four genera described, on the other hand, constitute a very natural group (the
Pelecanidae), capable of being characterized by a number of common characters drawn
from the postcranial part of the axial skeleton, many of which at the same time will
probably serve to mark it off from all other groups of birds.

Fregata and Phaeton agreed to differ from the Pelecanidae as follows : — The cervical
vertebra are only either twelve or thirteen in number, instead of from fifteen to
eighteen ; the cervical and cervico-dorsal vertebra together are only fifteen, instead of
from seventeen to twenty, as in the Pelecanidae. There is no distinct sacral vertebra at
all, or there is only a rudiment of sacral transverse processes.

In Fregata and Phaeton, again, the lateral acetabular fossa is enormous, instead of
being moderate or small as in the Pelecanidae. The sacro-sciatic foramen is very small
and short. In the Pelecanidae, except Phalacrocorax, there are always haemal arches to
some of the vertebrae, and in the last-named genus such arches are at least nearly
completed ; but in Fregata and Phaeton not only are there none, but no tendency to
form haemal arches is exhibited. In the two genera just named the dorsum of the
postacetabular part of the ilium is broad and dorsally convex, arching backwards and
downwards in a way not found in any of the Pelecanidae. In that family there is
always a marked interval between the lumbar and caudal vertebrae, through the non-
development of the parapophyses of some of the vertebrae. In Fregata (though not in

AXIAL SKELETON OP THE PELECANID^E. 365

Phaeton), on the other hand, there is no such interruption, the transverse processes
being continuously developed throughout these vertebrae.

But the most marked distinction obtains with respect to the cervical vertebras. In
all the Pelecanidae a sudden change takes place in the form of the cervical vertebras,
no vertebra being pressed back at its preaxial end before the eighth or ninth, while the
seventh or eighth exhibits, for the first time, serially, a postaxiad forking of its neural
arch. In Fregata and Phaeton, on the other hand, the change in the shape of the
vertebrae is gradual, and not sudden, the fifth vertebra being already pressed back post-
axiad at its preaxial end, while the same vertebra, or even the fourth, has already its
neural arch forking postaxiad, and therefore with a deeply concave postaxial margin to
its neural arch. Again, in all the Pelecanidee the two ilia meet medianly in front of
the acetabula and develop a median dorsal ridge, while in Fregata they do not nearly
meet together medianly in that region at all. In Phaeton, though they may appear so
to meet, yet they do not do so really, but each ilium has its mediad marginal ridge
distant from that of its fellow of the opposite side, though the membrane intervening
between the two may become ossified, and so produce a more or less deceptive appear-
ance of similarity to the structure of the Pelecanidae in this respect. Finally, in the
Pelecanidae there is but a single lateral xiphoid process on each side, while in Phaeton
there are two on each side, in addition to the median xiphoid process.

For all these reasons I think it better to keep the genera Fregata and Phaeton apart,
and, confining myself here to noticing their great distinctness (with respect to the axial
skeleton) from the Pelecanidse, to reserve any possible consideration of their positive
affinities till I come to treat of such group as may appear to exhibit similar cha-
racters.

Dr. Brandt, in his paper before referred to, considers the resemblances and affinities
of the Steganopodes amongst themselves and with other bird groups. But he bases his
estimates upon comparisons of other parts of their anatomy, scarcely making use of
characters drawn from the postcranial part of the axial skeleton. It seems to me
that though the four genera described form together a very natural group, yet it is
difficult to unite together any two of them to the exclusion of the others. Though in
some respects Sula resembles Pelecanus more than the other genera do, yet in other
respects Plotus and Pelecanus are most allied; and while Sula and Phalacrocorax
might from some axial characters be associated together, yet in others (e. g. in the
number of vertebrae and in the fact that it is the ninth from which its neighbours bend
dorsad) Phalacrocorax and Plotus may claim kinship. On the whole Plotus is the most
exceptional and differentiated type, and should therefore, I think, form one end of the
series, which may be begun with Pelecanus, which in some points, at least, appears the
least differentiated and most generalized form.

The characters of the four genera, and of the family they compose, may stand as
follows : —

3D2

366 PKOF. ST, GEORGE MIYART ON THE

CHARACTERS OF THE GENERA.

PELECANUS.

Cervical vertebrae 16, cervico-dorsal 1, together 17 ; dorsal 5, lumbar 3 or 4, lumbo-
sacral 3, sacral 3, sacro-caudal 4, caudal without pygostyle 5 or 6 : total 40-42. Vertebral
ribs 6, sternal ribs 5. Vertebrae generally swollen and very pneumatic ; no long, free
styloid processes, such processes merged in lateral subcentral plates ; anterior cervical
vertebrae elongated ; seventh and ninth bend dorsad from eighth ; median hypapophyses
conspicuous only in fifteenth, sixteenth, and seventeenth vertebrae, and not present in
lumbo-sacral or posterior dorsal regions ; a haemal arch to vertebrae from eighth to four-
teenth inclusive ; faint lateral ridges beneath centra of fifteenth, sixteenth and seven-
teenth vertebrae ; ridges and processes generally obtuse ; metapophyses moderate, rather
small. Atlas with an odontoid notch and minute hypapophyses. Hypapophyses of axis
very small; its hyperapophyses very small; a lateral foramen leading into centrum.
Third vertebra with a most rudimentary hypapophysis ; lateral vertebral canal long and
narrow ; hyperapophyses very small ; no marked interzygapophysial ridge to third ver-
tebra. Fifth vertebra with a median subcentral groove ; sixth vertebra with the same ;
postzygapophyses more postaxiad in seventh than in sixth vertebra; postaxial margin
of neural arch of seventh vertebra very concave, the first (preaxially) to be so. Eighth
vertebra the first vertebra pressed back preaxially, with a haemal arch, with no pro-
minent metapophyses, about equals the seventh in length ; styloid processes much as
in seventh ; prsezygapophyses not so preaxiad as centrum ; postzygapophyses slightly
more postaxiad of centrum than in seventh vertebra. Ninth vertebra but little more
pressed back preaxially than eighth ; neural spine not more developed in ninth than in
eighth vertebra ; a haemal arch present ; hyperapophyses in the form of two lateral
ridges. Neural spine most prominent in seventh and eighth vertebrae; metapophyses
not prominent. Tenth vertebra with a haemal arch. Postzygapophyses of eleventh ver-
tebra still reach the postaxial end of centrum ; in twelfth vertebra the postzygapophyses
for the first time fail to extend postaxiad as far as centrum. Thirteenth vertebra with a
haemal arch ; fourteenth vertebra with a complete, or nearly complete, haemal arch.
Fifteenth, sixteenth, and seventeenth vertebrae with a median hypapophysis ; postero-
external angle of centrum not drawn out into a triangular process. Nineteenth ver-
tebra without a hypapophysis, and not laterally compressed. Dorsa of tenth to four-
teenth cervical vertebrae flattened, but not presenting a longitudinal groove on each
side ; no notable lateral compression of the vertebral centra in any region, nor any
great ventral flattening; no vertebrae opisthocoelous ; parapophyses of second sacral
vertebra not stronger than those of first ; parapophyses of sacro-caudal vertebrae not
suddenly bent dorsad. Broad ilio-caudal process ; obturator-foramen oval, broad in
proportion to its length ; sacro-sciatic foramen not visible when the pelvis is viewed

AXIAL SKELETON OF THE PELECANID^E. 367

dorsally, owing to the breadth of the ilia; transverse diameter of pelvis between its
antitrochanteric processes nearly double the width of the most preaxial part of ilia ;
lateral acetabular fossa very large and quadrate ; a supraacetabular fossa ; pelvis of
moderate relative length ; postaxial half of external margin of ischium slightly convex ;
ventral surface of conjoined ischium and ilium extensive; ventral surface of ischium
slightly ridged ; isehium, external to obturator-foramen, narrow. Pygostyle curved,
convex dorsally, with its apex curved postaxiad and ventrad. The vertebral ribs have
generally uncinate processes, except the last; last sternal rib expanded proximally.
Sternum but little longer than broad ; lateral xiphoid processes narrow and pointed ;
pleurosteon wide, with five articular surfaces.

SULA.

Cervical vertebrae 15, cervico-dorsal 3, together 18 ; dorsal 6, lumbar 2-4, lumbo-
sacral 2 or 3, sacral 1 or 2, sacro-caudal 5 or 6 ; caudal, without pygostyle, 7 : total
43 or 44. Vertebral ribs 9, sternal ribs 6 or 7. Vertebrae generally but little swollen
or pneumatic ; styloid processes free, rather long and rather stout ; anterior cervical
vertebra? all short ; seventh and ninth bent dorsad from eighth ; hypapophyses of
first, second, third, and fourth vertebras large, and small ones to seventeenth and
eighteenth vertebree ; none present in posterior dorsal or lumbo-sacral region ; haemal
arches to vertebrae from the ninth or tenth to the thirteenth; no lateral ridges
beneath fifteenth to twenty-sixth vertebrae ; ridges and processes generally sharp ;
metapophyses relatively very large. Atlas with an odontoid foramen and bony hypa-
pophysis ; axis with very long hypapophysis, large hyperapophyses, and lateral foramen
leading into centrum ; third vertebra with very long hypapophysis, short and wide
lateral vertebral canal, much larger hyperapophyses than in Pelecanus, and very
marked interzygapophysial ridges ; fifth and sixth vertebra? with a median subcentral
groove ; postzygapophyses of seventh vertebra not more postaxiad than those of sixth ;
postaxial margin of neural arch of seventh vertebra not very concave, the first (pre-
axially) deeply concave behind being that of the eighth vertebra. Eighth vertebra
more pressed back preaxially than the seventh of Pelecanus, but less so than the
eighth of Pelecanus ; it is the first which is pressed back preaxially ; eighth vertebra
about as long as the seventh, with no haemal arch, but with prominent metapophyses ;
styloid processes rather larger than in the seventh vertebra, which has its praezygapophyses
slightly more preaxiad, and postzygapophyses decidedly more postaxiad than centrum.
Ninth vertebra much more pressed back preaxially than the eighth, and the first one
very much pressed back, and in its development intermediate between the eighth and
the ninth vertebrae of Pelecanus ; its neural spine much more developed than that of
the eighth vertebra generally, with no ha3mal arch ; its hyperapophyses two sharp
lateral processes ; neural spine most prominent in the ninth vertebra of all the cervical
vertebrae ; its metapophyses very long and sharp processes. Tenth vertebra with a

368 PROF. ST. GEORGE M1VAET ON THE

haemal arch. The postzygapophyses of the eleventh vertebra hardly reach, if they do
reach, the postaxial end of its centrum ; the postzygapophyses of the twelfth vertehra
fail decidedly to extend postaxiad as far as the centrum. Thirteenth vertebra with a
haemal arch ; but this part is suddenly wanting in fourteenth ; fifteenth, sixteenth, and
seventeenth vertebrae with only a slight median ridge instead of a marked hypapo-
physis; postero-external angles of centrum of eighteenth vertebra not drawn out into
triangular processes ; nineteenth vertebra without a hypapophysis, and only slightly
compressed laterally. Dorsa of tenth to fourteenth cervical vertebrae not even much
flattened, and therefore certainly not antero-posteriorly grooved. No notable lateral
compression of vertebral centra, but most developed in from twenty-first to twenty-fifth
vertebrae ; vertebrae fourteenth to seventeenth flattened beneath ; no vertebrae opistho-
ccelous ; parapophyses of second sacral vertebra very much stronger than those of the
first ; parapophyses of sacro-caudal vertebrae suddenly bent dorsad. Ilio-caudal pro-
cesses either truncated or pointed ; obturator-foramen oval, relatively longer than in
Pelecanus ; sacro-sciatic foramina visible when the pelvis is viewed dorsally ; transverse
diameter of pelvis between its antitrochanteric processes only slightly in excess of that
of most preaxial part of ilia ; lateral acetabular fossae considerable and triangular ; no
supraacetabular fossae ; pelvis of moderate length ; postaxial half of external margins
of ischium concave ; ventral surface of conj oined ischium and ilium very small ; ventral
surface of the ischium strongly ridged ; ischium, external to obturator-foramen, narrow.
Pygostyle long, prismatic, ventrally flattened, and with a strong dorsal ridge ; its ventral
margin strongly concave. All but the last two long vertebral ribs with uncinate
processes ; last sternal ribs not expanded proximally. Sternum more than twice as long
as broad ; lateral xiphoid processes somewhat spatulate ; pleurosteon more narrowed
dorso-ventrally than in Pelecanus, but with five articular surfaces.

PHALACROCORAX.

Cervical vertebrae 17 or 18, cervico-dorsal 3 or 2, together 20; dorsal 5, lumbar
4 or 5, lumbo-sacral 1-3, sacral 1 or 2, sacro-caudal 7 or 8 ; caudal, without pygostyle,
5-8 : total 46-49. Vertebral ribs 7 or 8, sternal ribs 5. Vertebrae generally but little
swollen or pneumatic ; styloid processes long and very slender ; anterior cervical
vertebrae rather shorter relatively than in Pelecanus ; eighth and tenth vertebrae bend
dorsad from ninth ; hypapophyses large in the first to fourth and fourteenth to twenty-
seventh vertebrae, large in dorsal aud lumbo-sacral regions ; neural spine most prominent
in the seventh, eighth, and ninth of all the cervical vertebrae ; no vertebra with a com-
plete haemal arch ; no lateral ridges beneath fifteenth to twenty-sixth vertebrae ; ridges
and processes generally very sharp ; metapophyses relatively very large and prominent
indeed. Atlas with an odontoid foramen, and with three hypapophyses ; axis with very
long hypapophysis, moderate hyperapophyses, and no lateral foramen leading into
centrum ; third vertebra with very long hypapophysis, wide and very short lateral

AXIAL SKELETON OF THE PELECANID^E. 369

canal, large and much pointed hyperapophyses, and with only slightly marked inter-
zygapophysial ridges ; fifth and sixth vertebrae not grooved beneath, but with a median
ridge ; postzygapophyses of seventh vertebra not more postaxiad than those of sixth ;
postaxial margin of neural arch of seventh vertebra not very concave, the first (pre-
axially) deeply concave behind being that of the eighth vertebra ; eighth vertebra
about as long as the seventh, with no haemal arch, but with very prominent and rugged
metapophyses, with styloid processes rather longer than in the seventh vertebra, and
with prsezygapophyses much more preaxiad, and postzygapophyses rather more post-
axiad than the centrum. Ninth vertebra, and the first one which is at all pressed
back preaxially ; it is like the eighth of Pelecanus ; its neural spine not so much
developed as that of the eighth vertebra, and always with no haemal arch ; its hy-
perapophyses two very sharp lateral processes ; its metapophyses sharp and long, but
not so much as those of Sula. Tenth vertebra with no haemal arch ; the post-
zygapophyses of the eleventh vertebra fail, for the first time, decidedly to reach the
postaxial end of centrum (therefore both zygapophyses fall short of antero-posterior
limits of centrum) ; thirteenth vertebra with no haemal arch, but the catapophyses almost
form one ; a long median hypapophysis suddenly appears on the fourteenth vertebra ;
fifteenth, sixteenth, and seventeenth vertebrae with large, median, plate-like hypapo-
physes ; postero-external angles of centrum of eighteenth vertebra drawn out into
triangular processes ; nineteenth vertebra with a long narrow hypapophysis, and with
its centrum laterally compressed ; dorsa of tenth to nineteenth cervical vertebrae not
even much flattened, and therefore certainly not antero-posteriorly grooved; great
lateral compression in centra of nineteenth to twenty-seventh vertebrae; sixteenth,
seventeenth, and eighteenth vertebrae expanded and flattened beneath in spite of the pre-
sence of a hypapophysis ; twentieth to twenty-fourth vertebrae opisthoccelous ; para-
pophyses of second sacral vertebra not much, if at all, stronger than those of the first ;
parapophyses of sacro-caudal vertebrae rather suddenly bent dorsad. Ilio-caudal pro-
cesses long and pointed ; obturator -foramen oval, still longer than in Sula ; sacro-
sciatic foramina visible when the pelvis is viewed dorsally ; transverse diameter ol
pelvis, between its antitrochanteric processes, falling short of that of the most preaxial
part of the ilia ; each lateral acetabular fossa small, and shaped as a parallelogram ; no
supraacetabular fossa ; pelvis very elongated ; postaxial half of external margin of
ischium strongly convex ; ventral surface of conjoined ischium and ilium extensive ;
ventral surface of ischium very strongly ridged; ischium, external to obturator-foramen,
narrow. Pygostyle bent up dorsad, and shaped like the claw of a feline mammal.
All the long vertebral ribs except the last with an uncinate process ; last sternal ribs not
expanded proximally. Length of sternum exceeding its breadth but little more than
in Pelecanus ; lateral xiphoid processes rather broad, but pointed ; pleurosteon much
narrower dorso-ventrally than in Pelecanus, and with only four articular surfaces.

370 PEOF. ST. GEOKGE MIVAET ON THE

PLOTUS.

Cervical vertebrae 18, cervico-dorsal 2, together 20 ; dorsal 5 or 6, lumbar 4, lumbo-
sacral 2 or 3, sacral 2, sacro-caudal 4 ; caudal, without pygostyle, 7 or 8 : total 45 or 46.
Vertebral ribs 7 or 8, sternal ribs 6. Vertebrae generally but little swollen or pneu-
matic ; styloid processes short in the more anterior vertebra?, but enormous in the eighth,
and long in the four or five succeeding vertebras ; anterior cervical vertebra? very long,
and increasing in length to the eighth ; eighth and tenth vertebra? bending dorsad
from ninth ; hypapophyses present in the first and in from the fifteenth to the twentieth
vertebra?; in twenty-third to twenty-sixth vertebra? a median subcentral process of
peculiar nature ; a complete ha?mal arch in ninth to fourteenth vertebra? ; strong lateral
ridges on centra of seventeenth to twenty-sixth vertebra?, bent down in twenty-second
to twenty-sixth ; ridges and processes generally rather sharper than in Sula ; meta-
pophyses as sharp as in Phalacrocorax, but smaller. Atlas with a very small
odontoid process and one median hypapophysis ; axis with the merest rudiment of a
hypapophysis, small hyperapophyses, and no lateral foramen leading into centrum ;
third and fourth vertebra? with no hypapophysis, no lateral canal, no interzygapophy-
sial ridge, and hyperapophyses slightly larger relatively than in Pelecanus; fifth and
sixth vertebra? medianly grooved beneath ; postzygapophyses of seventh vertebra not
more postaxiad than in the sixth ; postaxial margin of neural arch of seventh vertebra
not concave, this concavity first appearing in the eighth, and there very slight ; eighth
vertebra not pressed back preaxially (the ninth the first to be so pressed back), half as
long again as seventh, with no ha?mal arch, with rugged but not very prominent
metapophyses, and with styloid processes enormously longer than those of seventh
vertebra, with pra?zygapophyses much more preaxiad than centrum, and with post-
zygapophyses not quite so postaxiad as centrum, but very much less postaxiad than in
seventh vertebra ; neural spine prominent in seventh and eighth vertebra?. Ninth
vertebra pressed back to an enormous degree ; it is thus parallel to the eighth of Pele-
canus ; its neural spine not so much developed as that of eighth ; ninth vertebra with
a ha?mal arch, and the first to have one ; its hyperapophyses two lateral sharp pro-
cesses, which may be so continued as to form a dorsal arch on each side of the vertebra,
or else they help to give attachment to such a fibrous arch ; the metapophyses mere
ridges ; tenth vertebra with a ha?mal arch ; the postzygapophyses of the eleventh vertebra
do not quite reach the postaxial end of the centrum ; but they do not fail to do so for
the first time, because they so fail in the eighth (though not in the tenth) vertebra ;
postzygapophyses of twelfth vertebra fail very decidedly to reach the postaxial end
of the centrum ; thirteenth vertebra with a haemal arch ; fourteenth vertebra the last
with a ha?mal arch. A large, median, plate-like hypapophysis suddenly appears in the
fifteenth vertebra, and is still larger in the sixteenth ; it is small and postaxial in the

AXIAL SKELETON OF THE PELECANID^E. 371

seventeenth vertebra; postero-external angles of centrum of eighteenth vertebra ex-
tensive, and extending more preaxiad than in Phalacrocorax. Nineteenth vertebra with
a small hypapophysis ; centrum not at all laterally compressed, but ventrally flattened.
Dorsa of tenth to fourteenth vertebrae form a longitudinal groove; great lateral com-
pression in centra of twenty-second to twenty-sixth vertebrae ; seventeenth to twenty-
first vertebrae flattened beneath ; twenty-second to twenty-fifth vertebrae opisthoccelous ;
parapophyses of second sacral vertebra not stronger than those of first ; parapophyses
of sacro-caudal vertebrae rather suddenly bent dorsad. Ilio-caudal processes long and
pointed ; obturator-foramen shorter and rounder than in the three preceding genera ;
sacro-sciatic foramina hidden when the pelvis is viewed dorsally, because of the length
of the transverse processes of the sacro-caudal vertebrae, which extend the ilium so
much outwards on each side ; transverse diameter of pelvis between its antitrochanteric
processes falls short of that of the most preaxial part of the ilia ; lateral acetabular fossae
small and triangular ; no supraacetabular fossa ; pelvis rather short ; postaxial half of
external margin of ischium strongly convex ; ventral surface of conjoined ischium and
ilium rather extensive ; ventral surface of ischium very slightly ridged ; ischium, ex-
ternal to obturator-foramen, broad. Pygostyle much laterally compressed and dorso-
ventrally curved. All the long vertebral ribs with an uncinate process, except the
last rib ; last sternal rib not expanded proximally. Length of sternum exceeding its
breadth but little more than in Pelecanus ; lateral xiphoid processes narrow, long, and
pointed ; pleurosteon much narrower dor so- ventrally than in Pelecanus and with only
four articular surfaces.

CHARACTERS OF THE PELEOANID^E.

Cervical vertebrae 15-18, cervico-dorsal 1-3, together 17-20 ; dorsal 5 or 6, prelumbar
22—26, lumbar 3-5, together 8-10 ; cervical to lumbar inclusive 25-30, lumbo-sacral
1-3, lumbar and lumbo-sacral 5-7, sacral 1-3, sacro-caudal 4-8, caudal 5-8, sacral to
caudal inclusive but without pygostyle 13-18: total 40-49. Vertebral ribs 6-9; sacral
ribs 5-7. A sudden change of form arising in either eighth or ninth vertebra, such
vertebra being pressed back postaxiad at its preaxial end, and having both the vertebra
in front and behind it bent dorsad from it ; neural arch of seventh or eighth vertebra
suddenly deeply concave at its hinder margin, save in Plotus; median hypapophysis
always to seventeenth vertebra, and possibly present from fourteenth to twenty-
seventh vertebra; always a complete, or nearly complete, hremal arch to two or
three vertebrae, and generally complete from tenth to thirteenth vertebras, sometimes
from eighth to fourteenth ; metapophyses often long and pointed ; atlas generally
with an odontoid foramen, sometimes with an odontoid groove ; neural spine promi-
nent in seventh to ninth vertebra: ; postzygapophyses fail to attain the postaxial end
of the centrum, for the first time, in from eighth to twelfth vertebrae ; twentieth to

VOL. x. — PART vn. No. 8. — August 1st, 1878. 3s

372 PEOF. ST. GEOEGKE MIVAET ON THE

twenty-fifth vertebrae may be opisthoccelous ; a median snbvertebral process may be
formed, of bent-down lateral hypapophyses conjoined, as in Plotus; lateral acetabular
fossa subtriangular or quadrate, but never enormous in size ; supraacetabular fossa
present or absent; always a marked interval between lumbar and caudal vertebrae
through non-development of parapophyses ; sacral vertebrae distinct or not so. Pelvis
including from fifteen to seventeen vertebrae ; ilia not meeting at an acute angle ; no
conspicuous ilio-pectineal processes ; ilia not flattened against the postacetabular verte-
bree ; acetabula, preaxiad to the antero-posterior middle of the pelvis ; ischia and ilia
largely united postaxially; pelvis more vertically extended behind than in front of
acetabula; postacetabular part of pelvis broader transversely than the preacetabular
part; ilio-caudal spine present or absent; preacetabular parts of ilia meeting in a
median dorsal ridge; dorsum of postacetabular part of ilium not high, convex and
broad. Sternum twice as long as broad, or else but very little longer than broad ;
furcula ankylosed or not ; only one lateral xiphoid process on each side.

DESCRIPTION OF THE PLATES.

(In Plates LV., LVL, LVIL, & LVIIL, the following letters have throughout the

meanings here given.)

ac. Preaxial surface of centrum.

az. Praezygapophysis.

c, Catapophysis.

c'. Eidge extending postaxially from catapophysis.

eg. Groove internal to catapophysial ridge.

d. Diapophysis.

/. Foramen in catapophysial ridge.

g. Groove external to catapophysial ridge.

h. Surface for head of rib.

hp. Hyperapophysis.

Tip'. Small process, preaxial to hyperapophysis.

hy. Hypapophysis.

Tiy', Lateral parapophysial ridge.

t'z. Interzygapophysial ridge.

m. Metapophysis.

MS. Neural spine.

o. Odontoid process.

of. Foramen for odontoid process.

•p. Parapophysis.

AXIAL SKELETON OF THE PELECANID^. 373

p'. Parapophysial ridge.

pc. Postaxial surface of centrum.

pi. Pleurapophysial lamella.

ps. Styliform process.

pz. Postzygapophysis.

t. Surface for tubercle of rib.

v. Lateral canal.

z. Zygapophysis of atlas.

PLATE LV.

VERTEBRAE OF Pelecanus.

Fig. 1. Atlas vertebra, seen preaxially.
Fig. 2. Atlas vertebra, seen laterally.
Fig. 3. Axis vertebra, seen laterally.
Fig. 4. Third vertebra, seen laterally.
Fig. 5. Fifth vertebra, seen ventrally.
Fig. 6. Sixth vertebra, seen laterally.
Fig. 7. Sixth vertebra, seen dorsally.
Fig. 8. Sixth vertebra, seen ventrally.
Fig. 9. Seventh vertebra, seen laterally.
Fig. 10. Seventh vertebra, seen dorsally.
Fig. 11. Seventh vertebra, seen ventrally,
Fig. 12. Seventh vertebra, seen preaxially.
Fig. 13. Eighth vertebra, seen laterally.
Fig. 14. Eighth vertebra, seen dorsally.
Fig. 15. Eighth vertebra, seen ventrally.
Fig. 16. Eighth vertebra, seen preaxially.
Fig. 17. Eighth vertebra, seen postaxially.
Fig. 18. Ninth vertebra, seen laterally.
Fig. 19. Ninth vertebra, seen dorsally.
Fig. 20. Ninth vertebra, seen ventrally.
Fig. 21. Ninth vertebra, seen preaxially.
Fig. 22. Thirteenth vertebra, seen dorsally.
Fig. 23. Thirteenth vertebra, seen ventrally.
Fig. 24. Thirteenth vertebra, seen preaxially.
Fig. 25. Fifteenth vertebra, seen laterally.
Fig. 26. Fifteenth vertebra, seen dorsally.
Fig. 27. Fifteenth vertebra, seen ventrally.
Fig. 28. Fifteenth vertebra, seen preaxially.
Fig. 29. Fifteenth vertebra, seen postaxially.

3E 2

374 PBO.F. ST. GEOKGE MIVAKT ON THE

PLATE LVI.

VERTEBRA OF Sula.

Fig. 1. Atlas vertebra, seen preaxially.
Fig. 2. Atlas vertebra, seen laterally.
Fig. 3. Axis vertebra, seen laterally.
Fig. 4. Third vertebra, seen laterally.
Fig. 5. Fifth vertebra, seen ventrally.
Fig. 6. Sixth vertebra, seen laterally.
Fig. 7. Sixth vertebra, seen dorsally.
Fig. 8. Sixth vertebra, seen ventrally.
Fig. 9. Seventh vertebra, seen laterally.
Fig. 10. Seventh vertebra, seen dorsally.
Fig. 11. Seventh vertebra, seen ventrally.
Fig. 12. Seventh vertebra, seen preaxially.
Fig. 13. Eighth vertebra, seen laterally.
Fig. 14. Eighth vertebra, seen dorsally.
Fig. 15. Eighth vertebra, seen ventrally.
Fig. 16. Eighth vertebra, seen preaxially.
Fig. 17. Eighth vertebra, seen postaxially.
Fig. 18. Ninth vertebra, seen laterally.
Fig. 19. Ninth vertebra, seen dorsally.
Fig. 20. Ninth vertebra, seen ventrally.
Fig. 21. Ninth vertebra, seen preaxially.
Fig. 22. Tenth vertebra, seen laterally.
Fig. 23. Tenth vertebra, seen dorsally.
Fig. 24. Tenth vertebra, seen preaxially,
Fig. 25. Fifteenth vertebra, seen dorsally.
Fig. 26. Fifteenth vertebra, seen ventrally.
Fig. 27. Fifteenth vertebra, seen laterally.
Fig. 28. Fifteenth vertebra, seen preaxially.

PLATE LVII.

VERTEBRAE OP Phalacrocorax,

Fig. 1. Atlas vertebra, seen preaxially ,"1

Fig. 2. Atlas vertebra, seen laterally, >once and a half the natural size.

Fig. 3. Atlas vertebra, seen ventrally. J

Fig. 4. Axis vertebra, seen laterally, natural size.

AXIAL SKELETON OF THE PELECAJSIIIXS;. 375

Fig. 5. Axis vertebra, seen ventrally.
Fig. 6. Third vertebra, seen laterally.
Fig. 7. Third vertebra, seen ventrally.
Fig. 8. Seventh vertebra, seen laterally.
Fig. 9. Seventh vertebra, seen dorsally.
Fig. 10. Seventh vertebra, seen ventrally.
Fig. 11. Seventh vertebra, seen preaxially,
Fig. 12. Eighth vertebra, seen laterally,
Fig. 13. Eighth vertebra, seen dorsally,
Fig. 14. Eighth vertebra, seen ventrally.
Fig. 15. Eighth vertebra, seen preaxially.
Fig. 16. Eighth vertebra, seen postaxially,
Fig. 17. Ninth vertebra, seen laterally.
Fig. 18. Ninth vertebra, seen dorsally.
Fig. 19. Ninth vertebra, seen ventrally.
Fig. 20. Ninth vertebra, seen preaxially.
Fig. 21. Tenth vertebra, seen laterally.
Fig. 22. Tenth vertebra, seen dorsally.
Fig. 23. Tenth vertebra, seen preaxially.
Fig. 24. Twelfth vertebra, seen laterally.
Fig. 25. Fifteenth vertebra, seen laterally.
Fig. 26. Fifteenth vertebra, seen dorsally.
Fig. 27. Fifteenth vertebra, seen ventrally.
Fig. 28. Fifteenth vertebra, seen preaxially,
Fig. 29. Fifteenth vertebra, seen postaxially.
Fig. 30. Eighteenth vertebra, seen laterally,
Fig. 31. Nineteenth vertebra, seen laterally.
Fig. 32. Nineteenth vertebra, seen preaxially.
Fig. 33. Nineteenth vertebra, seen postaxially,
Fig. 34. Twentieth vertebra, seen laterally.
Fig. 35. Twentieth vertebra, seen preaxially.
Fig. 36. Twentieth vertebra, seen postaxially.
Fig. 37. Twenty-first vertebra, seen laterally.
Fig. 38. Twenty-first vertebra, seen preaxially.
Fig. 39. Twenty-first vertebra, seen postaxially.
Fig. 40. Twenty-second vertebra, seen laterally.
Fig. 41. Twenty-second vertebra, seen preaxially.
Fig, 42. Twenty-second vertebra, seen postaxially.
Fig. 43. Twenty-third vertebra, seen laterally.
Fig. 44. Twenty-third vertebra, seen preaxially.
Fig. 45. Twenty-third vertebra, seen postaxially,

376 PEOF. ST. GEOKGE MIVAET ON THE

PLATE LVIII.

VERTEBRA OF Plotus.

Fig. 1. Atlas vertebra, seen preaxially,) . ,

_. 5 >twice the natural size.

Fig. 2. Atlas vertebra, seen laterally, )

Fig. 3. Axis vertebra, seen laterally, the size of nature.

Fig. 4. Axis vertebra, seen ventrally.

Fig. 5. Third vertebra, seen laterally.

Fig. 6. Third vertebra, seen ventrally.

Fig. 7. Fourth vertebra, seen laterally.

Fig. 8. Fourth vertebra, seen ventrally.

Fig. 9. Fifth vertebra, seen laterally.

Fig. 10. Fifth vertebra, seen dorsally.

Fig. 11. Fifth vertebra, seen ventrally.

Fig. 12. Sixth vertebra, seen laterally.

Fig. 13. Sixth vertebra, seen dorsally.

Fig. 14. Sixth vertebra, seen ventrally.

Fig. 15. Seventh vertebra, seen laterally.

Fig. 16. Seventh vertebra, seen dorsally.

Fig. 17. Seventh vertebra, seen ventrally.

Fig. 18. Eighth vertebra, seen laterally.

Fig. 19. Eighth vertebra, seen dorsally.

Fig. 20. Eighth vertebra, seen ventrally.

Fig. 21. Eighth vertebra, seen preaxially.

Fig. 22. Eighth vertebra, seen postaxially.

Fig. 23. Ninth vertebra, seen laterally.

Fig. 24. Ninth vertebra, seen dorsally.

Fig. 25. Ninth vertebra, seen ventrally.

Fig. 26. Ninth vertebra, seen preaxially.

Fig. 27. Ninth vertebra, seen postaxially.

Fig. 28. Tenth vertebra, seen laterally.

Fig. 29. Tenth vertebra, seen dorsally.

Fig. 30. Tenth vertebra, seen preaxially.

Fig. 31. Eleventh vertebra, seen laterally.

Fig. 32. Eleventh vertebra, seen ventrally.

Fig. 33. Twelfth vertebra, seen laterally.

Fig. 34. Fifteenth vertebra, seen laterally.

Fig. 35. Fifteenth vertebra, seen dorsally.

Fig. 36. Fifteenth vertebra, seen preaxially.

AXIAL SKELETON OF THE PELECANLD^E. 377

Fig. 37. Eighteenth vertebra, seen laterally.

Fig. 38. Eighteenth vertebra, seen ventrally.

Fig. 39. Nineteenth vertebra, seen laterally.

Fig. 40. Nineteenth vertebra, seen ventrally.

Fig. 41. Twentieth vertebra, seen ventrally.

Fig. 42. Twentieth vertebra, seen preaxially.

Fig. 43. Twentieth vertebra, seen postaxially.

Fig. 44. Twenty-second vertebra, seen postaxially.

Fig. 45. Twenty-third vertebra, seen preaxially.

Fig. 46. Twenty-first, twenty-second, twenty-third, and twenty-fourth vertebrae

together, seen ventrally.

Fig. 47. Caudal vertebrae and pygostyle.

(In Plates LIX., LX., & LXI., the following letters have throughout the

meanings here given.)

a. Sternal angle postaxial to pleurosteou (Plotut).

at. Antitrochanterie process.

e. Coracoid groove of sternum,

ca. Costal angle of sternum.

cl. Boot of furculum.

/. Lateral acetabular fossa.

f. Supraacetabular fossa (Pelecanus).

g. Groove on ventral surface of ischium.

gl. Gluteal ridge.

h. Surface for head of rib.

i. Ischium.

it. Hium.

il'. PostacetaLular part of ilium.

k. Keel of sternum.

Ic. Ilio-caudal spine.

lp. Ilio-pectineal eminence.

Ir. Eidge postaxial to posttrochanterie process (Plotut).

LK. Lateral xiphoid.

nix. Median xiphoid.

p. Pubis.

ps. Ventral process of ischium.

*1

s2. V Sacral foramina.

«».J

st. Posttrochanteric process.

t. Surface for tubercle of rib.

378 PEOF. ST. GEOEGE MIVAET ON THE PELECANIDJS.

PLATE LIX.

Fig. 1. Dorsal view of pelvis of Pelecanus.

Fig. 2. Lateral view of pelvis of Pelecanus.

Fig. 3. Ventral view of pelvis of Pelecanus.

Fig. 4. Lateral view of sternum of Pelecanus.

Fig. 5. Ventral view of sternum of Pelecanus.

Fig. 6. Caudal vertebrae and pygostyle of Pelecanus.

Fig. 7. Pygostyle of Sula.

PLATE LX.

Fig. 1. Dorsal view of pelvis of Sula.

Fig. 2. Lateral view of pelvis of Sula.

Fig. 3. Ventral view of pelvis of Sula.

Fig. 4. Lateral view of sternum of Sula. (The sixth sternal

rib is not represented.)
Fig. 5. Ventral view of sternum of Sula.
Fig. 6. Lateral view of pelvis of Phalacrocorax.

PLATE LXI.

Fig. 1. Ventral view of pelvis of Phalaerocorax.

Fig. 2. Dorsal view of pelvis of Phalacrocorax.

Fig. 3. Lateral view of sternum of Phalacrocorax.

Fig. 4. Ventral view of sternum of Phalacrocorax.

Fig. 5. Some caudal vertebrae and pygostyle of Phalacrocorax.

Fig. 6. Lateral view of pelvis of Plotus.

Fig. 7. Dorsal view of pelvis of Plotus.

Fig. 8. Ventral view of pelvis of Plotus.

Fig. 9. Lateral view of sternum of Plotus.

Fig. 10. Ventral view of sternum of Plotus.

W.HWeslcy ad nat del

AXIAL SKELETON OF THE PELE

Hanhart imp

WHWesleyadnat.de!

AXIAL SKELETON OF THE PELECANIDJi.

X H 71..S.

WHWe.iley adult del.

AXTAT, ,SKF,T,T\T01\T OF THE PELE C ANID^E

Hanhart

imp.

12.

21.

22.

VniW..s;=v alnat del

AXIAL SKELETON OF THE

Hanhart

•

' J
w

. \

o

' !

E-H

.-1

A <

u / A'UML

r

o
W
h-1

w

Pi

w

I'
o

W

H-I

w

CO

H-l

I — I

X

w
i-q
W

PH'

H
W

E-i

o

'H

H
^
W
M
to

H

•j

14 DAY USE

RETURN TO DESK FROM WHICH BORROWED

SCIENCES LIBRAKi

This book is due ou the last date stamped below, or
on the date to which renewed.
Renewed books are subject to immediate recall.

LD 21-50m-4,'63 General Library
(D6471«10)416 University of California
Berkeley

U.C.BERKELEY LIBRARIES